Method for at least partially automatically controlling a container treatment device for treating containers and container treatment device

Abstract

Method for the at least partially automatic, preferably fully automatic, control of a container treatment device (1) for treating containers (10), which device has a transport device (2) for guiding containers (10) along a predetermined transport path, as a function of an occupancy level of at least one transport area (20) along the transport path, wherein image data (E1, E2) are generated by means of at least one image capture device (4), which image data depict containers (10) located in the transport area (20) and on the basis of which image data an image evaluation device (6) carries out an evaluation, in particular a real-time evaluation, to determine an occupancy level variable characteristic of the occupancy level of the transport area (20). According to the invention, the image evaluation device (6) carries out an at least section-wise segmentation of the image data (E1, E2) to determine the occupancy level variable.

Description

The present invention relates to a method for at least partially automatically controlling a container treatment device for treating containers and to a container treatment device.The present invention further relates to a method and a training data generation device for automatically generating a training dataset for training, in particular for retraining, a container recognition model of machine learning of a container treatment device, in which containers can be or are transported along a predefined transport path.In this case, image data can be generated or are generated by means of an image capturing device, which image data are characteristic of a transport region along the transport path. The transport region is preferably a region within which the containers located therein are to be detected. The image data can be supplied to the container recognition model for performing an evaluation, in particular a real-time evaluation, with respect to the containers located in the transport region as input variables and are in particular supplied to the latter.The containers are preferably plastic containers (in particular PET containers), containers whose main constituent consists of pulp and / or glass containers and / or cans. The containers can be containers from the beverage and / or food and / or cosmetics industry. For example, these cans or bottles, such as glass bottles, pulp bottles and plastic bottles.The containers (to be recognized) can be fully formed containers and / or not yet fully formed containers such as preforms. Furthermore, they can be empty containers or containers already filled with a product. Furthermore, the containers to be identified can be provided with closures or not with closures.Different objects can be recognized with the aid of neural networks. The present invention relates in particular to the recognition of containers or of parts of containers (cover, bottle mouth etc.).In order for the neural network to be able to recognize these containers, it must be trained. Training data is required for this purpose. The training data consist in particular of labeled images.The objects to be detected later can be marked with so-called "region of interest" (ROl). These can preferably be rectangles, circles, or else a polygon. If a plurality of relevant objects are present in an image, a plurality of ROls must also be created. This process is referred to as labeling. The algorithm learns during later training to identify these ROls in other situations. The labeling process takes place completely manually or semi-automatically in the prior art.In manual labeling, the images are provided with annotation data manually by a user. There are various approaches to semi-automated labeling. Most approaches use a coherent sequence of images (video). The user labels a part of the images manually and these annotation data are extrapolated or interpolated to the following images with the aid of an algorithm. It is also possible to use an already existing neural network for labeling. In this variant, the user must decide whether the objects have been marked correctly. Regardless of the variants, manual intervention by a user is always necessary.Experience from the methods known in the prior art shows that a trained neural network cannot recognize every container. Retraining will always be necessary for the respective situation. Thus, labeled data is needed for training. Manual labeling takes a very long time and must be carried out by a worker.In the field of filling systems, the traditional method for evaluating the occupancy level is mainly based on the use of sensors. These sensors are strategic along transport and buffer segments and function as observers responsible for monitoring and regulating the individual machine operations. They measure and collect important occupancy data, and these measurements allow precise control over machine performance, taking into account the speed of the transporters. Therefore, the more accurate the measurements are, the more precise the control over the machine's performance and the transporter's speeds is. Accurately measured occupancy levels may indicate whether a machine should reduce its performance or a transporter should increase its speed to prevent jamming.The sensors currently used for evaluating the occupancy level in the area of the filling systems are congestion switches. However, these backup switches only detect whether a section is completely filled. The occupancy level from the congestion switch is thus 100%. If the congestion switch is not busy, the occupancy level is between 0 and <100%. However, a more accurate occupancy level cannot be determined with such congestion switches.However, a considerable disadvantage of this traditional approach is the imprecision of these sensors. This imprecision can be particularly problematic if it is a matter of transporters or buffer systems that consist of several tracks or have complex geometries. As a result, the data provided by these sensors often does not provide an accurate representation of the actual occupancy levels, which presents challenges in maintaining precise control over the transport speeds / machine powers in the system.The effects of inaccurate occupancy data go beyond pure data inconsistencies. They directly affect the performance of the machines and the production lines within the filling system. Machines that are regulated based on erroneous occupancy information may operate inefficiently, which may result in production bottleneckes and reduced throughput. Moreover, these inaccurate data, congestion, and transport disturbances may occur more frequently because the control of the transport speeds / machine powers is based on these inaccurate data. A traffic jam during transport not only impairs production, but also carries the risk of damage to the transported products.EP 2 132 129 B1 discloses a method for monitoring, controlling and optimizing filling systems in which finely graduated occupancy levels are determined for each individual buffer section, wherein a recorded image is analyzed to determine these finely graduated occupancy levels as to which number of bottles or cans is located on the buffer section at the time of recording, wherein each individual bottle or can is recognized as an object of the type, i.e. as such.EP 300 523 1 B1 discloses a method for counting objects on a conveyor belt. Here, tracking a position of an extracted characteristic of each object in the image data in frames is proposed by extracting the characteristic in at least one additional subsequent frame recorded by the same camera.The present invention is based on the object of overcoming the disadvantages known from the prior art and of providing and providing a precise, rapid and robust method for controlling a container treatment device as a function of a degree of occupancy of a transport region and for ascertaining a degree of occupancy of the transport region and a corresponding container treatment device. A further object is to provide a method and a training data generation device for generating training data for training, in particular for retraining, a container recognition model of machine learning of a container treatment device, which can carry out the marking or labeling of the training images as far as possible without intervention by a user. A further object is to provide a precise, rapid and robust method for detecting containers located in a transport region of a container treatment device and a corresponding container treatment device.The object is achieved according to the invention by the subject matter of the independent claims. Advantageous embodiments and developments of the invention are the subject matter of the dependent claims.In a method according to the invention for at least partially automatic, preferably fully automatic, control of a container treatment apparatus for treating containers, the container treatment apparatus has a transport device for guiding containers along a predefined transport path. The control of the container treatment device is effected in this case as a function of a degree of occupancy of at least one (predefined and / or predefinable) transport region (of the transport device) along the transport path.In this case, (in particular in a working mode of the container treatment apparatus) image data (preferably at a common recording time) are generated by means of at least one image recording device (preferably with a plurality of image recording devices), said image data imaging containers located in the transport region and on the basis of which an, in particular processor-based, image evaluation device carries out an evaluation, in particular a real-time evaluation, for determining an occupancy level variable characteristic of the occupancy level of the transport region.The image data are preferably generated in the working mode during continuous transport of the containers by the transport device, i.e. without the image data collection by the image acquisition device causing and / or causing an influence on the transport speed of the containers and / or correlating with such.In particular, the occupancy level variable is for the containers located in the transport area, preferably for the number of containers located in the transport area and / or for the area fraction of the transport area occupied by the containers located in the transport area.The transport area is preferably arranged in a capture area of the at least one image capture device such that the entire (predefined and / or predefinable) transport area and / or containers located in the entire (predefined and / or predefinable) transport area can be captured by the at least one image capture device. The image data generated (with the at least one image capturing device) is preferably characteristic of the containers located in the transport region (at the recording time).The image capturing device is preferably an image capturing device of the container treatment apparatus. The image evaluation device is preferably an image evaluation device of the container treatment device.The image evaluation device preferably carries out the evaluation of the image data to be evaluated by means of a (in particular computer-implemented) container evaluation model.The image capturing device (or the plurality of image capturing devices) can be an image capturing device such as a camera (preferably black and white and / or colored), a CMOS sensor (CMOS Abk. for Complementary metal-oxide-Semiconductor), a CCD sensor, a 3D sensor, an X-ray-based image capturing device, an optical element, a thermal imaging camera, a stereo camera, a LIDAR camera and the like, and combinations thereof.The image data (generated by the image acquisition device and / or to be transmitted to and / or supplied to the image evaluation device) is preferably two-dimensional (spatially resolved) image data. The image data which can be transmitted (or are transmitted) and / or are supplied or are supplied to the image evaluation device for performing an evaluation preferably comprise no depth information (measured directly or directly by the image detection device), that is to say in particular no (depth and / or distance) measured values measured in the direction of the recording direction of the image detection device, which are characteristic of a distance of the image detection device from the objects imaged in the image data. In other words, the image capturing device preferably does not generate (and / or capture) a measured value which is characteristic solely of a distance and / or a (relative) position of the image capturing device with respect to an imaged object. Such a distance and / or a position of the image capturing device relative to the objects imaged by it could be determined from image data generated by image capturing device(s) at mutually different positions (for example via stereo recordings and / or a LIDAR camera). It is preferably possible, as described in more detail in a subsequent section within the scope of the method for automatically generating a training dataset for training a container recognition model of machine learning of a container treatment apparatus (to which reference is made in particular here), to calculate back to the spatial coordinates with a 2D image (recorded by the image capturing device) (if, for example, the diameter of the recognized or of the transported and / or of the container to be treated is known or alternatively a reference surface or reference line with a known geometric extent is captured by the image capturing device).Preferably, the image capturing device records a color image or a color video sequence or a color image sequence (for ascertaining and / or generating the image data). However, it is also conceivable that the recorded and / or generated and / or determined image data is a grayscale image or grayscale images. In other words, it is conceivable for achromatic image data to be transmitted (or supplied as input variables) to the image evaluation device or the image evaluation model. In this way, a larger transport region of the container treatment device can advantageously be detected or mapped and evaluated in the image data.The image capturing device (or the plurality of image capturing devices) is preferably suitable and determined for recording (static) (single) images and / or moving images or image sequences (or video sequences) or is used for this purpose.The image data to be supplied (generated by the image capturing device) to the image evaluation device (or to the image evaluation model) can be image sequences and / or individual images and / or image data recorded (essentially) at a single recording time or (essentially) simultaneously recorded or captured.It is conceivable that the image data (on the basis of which the image evaluation device performs the evaluation) (to be supplied to and / or transmitted to the image evaluation model) is image data determined from more than one individual image. For example, two (mutually different) image capturing devices can each record an image (preferably substantially simultaneously). Preferably, an image composed of a plurality of images and / or image data generated from a plurality of images is supplied to the image evaluation device and / or the image evaluation model.It is additionally or alternatively conceivable that the image data (on the basis of which the image evaluation device carries out the evaluation) is generated on the basis of a plurality of individual images (recorded and / or recorded by the at least one image recording device), in particular recorded and / or recorded at different recording times). Individual images that could also be used are, for example, individual frames of a video sequence (for example recorded by the at least one image capturing device). For example, a plurality of frames (recorded successively in time for instance) can be averaged and / or added to generate the image data (on the basis of which the image evaluation device performs the evaluation).The preferably multiple image acquisition devices are preferably synchronized and / or synchronizable with one another for recording or generating the image data. The plurality of image capturing devices preferably cover the entire transport region with their capturing regions. However, it is also conceivable for the plurality of image capturing devices to record the image data at different recording times and for the image data to be processed (for example depending on the transport speed of the transport device and / or relative arrangement of the image capturing devices) in such a way that image data are obtained therefrom which are characteristic of the occupancy level of the transport area present at the same time and / or a location and / or an orientation of the containers located in the transport area.The (at least one) image capturing device can preferably be arranged above the transport region. It is conceivable that the image capturing device is arranged obliquely above the transport region and thus captures in particular image data from a capturing direction which encloses an angle different from 90° with a transport plane. The transport plane can be, for example, the contact surface with the containers provided by the transport device, for example a transport track, during the transport of the containers. If the containers are transported, for example, (upright), the transport plane could be that surface of the transport device which can contact the container bases during their transport.The image capturing device can be arranged symmetrically and / or centrally with respect to an extent of the transport region in a width direction, wherein the width direction is in particular a direction perpendicular to the transport direction or to the transport path. However, a lateral arrangement of the image capturing device with respect to the (middle) transport path and / or the transport region is also conceivable.The image capturing device is preferably arranged vertically above the transport region (at least in sections and preferably completely). In particular, the acquisition direction of the image acquisition device encloses a 90° angle with at least one section of the transport area and preferably with the entire transport area.The container treatment apparatus (in particular each transport device of its own) preferably has exactly one image capturing device for capturing image data with respect to containers located in transport sections.Preferably, pre-processed image data are supplied to the image evaluation device and / or the image evaluation model. Preprocessing steps of the image data preferably comprise cropping (for example to the transport region or to the image data points imaging the transport region), sharpening and / or changing the brightness (brightness). Preferably, there is no change in the color range of the image data, in particular no conversion of a color image into a grayscale image. However, it would also be conceivable for the color values of the image data to be changed or for the color range of the image data to be changed.However, it is also conceivable for the image evaluation device to perform preprocessing of the image data and / or preprocessing steps (for example mentioned above) (individually or in combination).According to the invention, the image evaluation device performs at least partially segmenting of the image data in order to determine the occupancy level variable. The image data in this case are, in particular, the image data to be evaluated by the image evaluation device (and / or supplied to the image evaluation model, in particular as an input variable). However, it is also conceivable for the image evaluation device to perform (possibly as described above) preprocessing of the image data to be evaluated (by it) before a segmentation of the image data obtained by the preprocessing is carried out.The image evaluation model is in particular a (preferably semantic) segmentation model to which image data can be supplied as input variables and by means of which a segmentation of the image data supplied as input variables can be carried out.The determined occupancy level variable preferably serves as a control variable for the container treatment device, for example for (automatically) carrying out a treatment function on at least one and / or on a plurality of containers. It is also conceivable that the at least one occupancy level variable is used as a control variable for controlling and / or regulating the container flow and / or a container throughput through the container treatment device and / or a transport speed and / or a container feed and / or container discharge.In particular, performing a segmentation is understood to mean a process of image evaluation and / or image processing (which is implemented by a computer), in particular carried out by the image evaluation device, in which the image data (which is to be evaluated by the image evaluation device and / or is to be segmented) supplied as input variables to the image evaluation model, in particular the segmentation model, are divided into segments (into meaningful image parts). The segments preferably have a (particularly preferably contiguous) set of data points which is distinguished by a specific or predefined and / or predefinable property or attribute or which satisfies a predefined and / or predefinable specific relation (for example, gray values and / or color values are in a predefined color range).Preferably, the segmentation or the division of the image data into segments takes place with respect to the containers (to be treated). Preferably, the segmentation of the image data or the division of the image data takes place as a function of a judgment as to whether these (at least partially) image a container (or whether these do not (at least partially) image a container).Preferably, the segmentation of the image data and / or the division of the image data into segments with respect to the containers (to be treated) takes place as a function of a predefined and / or predefinable detection range on the containers (to be treated).The term recognition region on the containers (to be treated) is understood here in particular to mean that the recognition region can be a partial region of the container (for example the container top side and / or container bottom side) or a region of an element which is arranged on the container (preferably rotationally fixed and / or translation-fixed) (such as, for example, a closure and / or equipment of the container, such as, for example, a label). This offers the advantage that a region which can be identified particularly well by optical image evaluation and / or by image processing (for example by comparatively high contrast and / or conspicuous color values with respect to the remaining regions of the container and / or with respect to objects surrounding the containers and / or the transport region, such as a (specular) railing), can be specified or selected as the recognition region.In a preferred method, the recognition region (of the containers to be treated) is selected from a group which comprises lids, closures, can top sides, can bottom side, container top side, container bottom side wall regions, a mouth region of the container to be recognized, equipment of a container (such as, for example, a label applied to the container, for example), a logo arranged on the container, a closure, a bundle of containers (entire bundles), and the like, and combinations and subareas thereof. The recognition region is preferably a partial surface of a container (depicted in a two-dimensional image). The detection region can be characteristic of an element (such as a label) arranged on the container and / or of a (partial) region of the container.For example, in a preferred method, the recognition area can be container covers, for example bottle covers.Preferably, no recognition of individual containers is carried out during the segmentation. In particular, for example during the segmentation, the object boundaries between adjacent containers are not determined. In other words, the segments obtained by the segmentation (for example in the image evaluation model, in particular in the segmentation model as output variables) indicate in particular no object boundaries and / or position and / or orientation of individual containers of a plurality of containers adjoining one another. In other words, a segment obtained by the segmentation, the image data points of which depict a multiplicity of containers adjoining one another (at least in pairs), is not characteristic of an object boundary or the object boundaries running between the containers adjoining one another and / or of an alignment and / or orientation of the containers.In particular, it is indistinguishable, for example solely on the basis of a segment obtained by the segmentation, which segments summarizes image data points which depict at least one container (located on the transport region and captured by the generation of the image data), whether exactly one container is located on the transport region in an orientation located on the transport region or whether two or more containers are instead located in its position in a vertical orientation (and altogether claim approximately a comparable support region of the transport region).In a preferred method, the occupancy level variable is determined on the basis of at least one segment obtained by the segmentation and preferably at least one plurality of segments obtained by the segmentation. This offers the advantage that a very rapid and robust determination of the occupancy level variable is possible as a result.The occupancy level variable is preferably determined on the basis of the (determined) pixel number or number of image data points of at least one segment and preferably those segments which are assigned to container-imaging image data points. For example, this number or a variable derived therefrom can be compared with (at least) one comparison variable. The comparison can be made, for example, by forming a ratio and / or a difference. In addition, the comparison result obtained from the comparison can be compared with a (predefined and / or predefinable) threshold value.The (at least one) comparison variable can be predefined and / or (for example, be predefinable by an operator).The (at least one) comparison variable can be determinable or determined at least partially automatically and preferably fully automatically (by the container treatment device). It is conceivable that an adjustment operation of the container treatment device provided for this purpose (described in more detail in the following) is provided, which preferably differs from an (intended) working operation of the container treatment device (with the highest possible production numbers or transport speeds) in which the containers are treated.The (at least one) comparison variable is preferably stored in the image evaluation device. The stored comparison variable can preferably be changed, wherein particularly preferably the change can be determined by an operator and / or automatically (by the container treatment device).It is conceivable that the number of image data points and / or a measure for an area depicted by the image data (for example, relative to a plane within which the transport area extends) is used as comparison variable, on the basis of which the segmentation is carried out. This comparison variable can serve as a reference variable for the number of image data points (or measure for the area imaged by the image data points) of the respective segments, so that objective evaluation is possible.In a further preferred method, the occupancy level variable is determined as a function of a transport area variable, which is characteristic for an occupancy area and / or for a maximum occupancy number of the transport area. This preferably enables a more precise evaluation of the image data, because any background image data points, i.e. image data points which do not depict the transport region (or containers located thereon) but rather a surrounding region surrounding the transport region, are not taken into account in the evaluation of the segments (or segment sizes thereof) obtained by the segmentation.Preferably (therefore) the transport area size is selected as (at least one) comparison variable.The maximum occupancy count of the transport area is to be understood in particular as the maximum count of the containers receivable in the transport area.A transport area size characteristic of a maximum occupancy number of the transport area can also be understood to mean, for example, that area (-on-) part of the total area of the transport area which, in the case of maximum occupancy, i.e. in the case of close packing of the containers, is occupied by the containers of a maximum number of containers that can be accommodated in the transport area. Such a transport area size could therefore indicate, for example, a number of image data points / pixels or be characteristic thereof, which in each case depict a container area of one of the containers when the transport area is occupied to a maximum extent (when the containers are packed tightly).The occupancy area is preferably understood to mean a size characteristic of the area of the transport area within which containers can be accommodated and / or transported and / or guided, or a size characteristic thereof. The occupancy area can thus represent a measure for the area of the real transport area. However, it would also be conceivable for the occupancy area to represent a measure for the (total) area of the transport area (depicted in the image data).The comparison variable and / or the transport region size can / can be independent of the type of containers (to be treated). If, for example, a transport belt serves as a transport device, the total area of the transport belt which lies in the transport area can be selected as the transport area size (or a corresponding size which is characteristic of this total area depicted in the image data). In particular, for example, containers of round cross section (such as cans or bottles known from the beverage industry) do not completely fill this transport region when the package is tight on the transport belt or in the transport region. Nevertheless, the total area of the conveyor belt can already provide a good approximation for a variable characteristic of a maximum occupancy number of the transport region (in particular for containers of known (cross-sectional) geometry).The above-mentioned threshold value can be dependent on the cross-sectional area occupied by the container (to be treated). Thus, a plurality of threshold values could be stored in the image evaluation device (for example in the form of a database), which are each assigned to a different container type.In this case, the cross-sectional area and / or cross-sectional geometry is considered to mean, in particular, a corresponding cross-sectional size with respect to a cross-sectional plane which is perpendicular to a detection direction of the (at least one) image detection device.The comparison variable and / or the transport region size can / can depend on the type of containers (to be treated) (in particular a cross-sectional area and / or a diameter). This can be the case, for example, if a transport area size characteristic of a maximum occupancy count is selected as transport area size, since here a denser or a less dense packing can be possible depending on the cross-sectional geometry.The container treatment device preferably has at least one human-machine interface (HMI) and / or an input device, via which (in particular by an operator of the container treatment device) the transport region size can be input (manually). It is also conceivable for the container treatment device to have a receiving device, by means of which the transport area size (for example from an, in particular cloud-based, (backend) server (for example from a manufacturer of the container treatment device)) can be transmitted to the container treatment device (for example via a wireless communication connection).It is also conceivable that image data and / or position data, which preferably indicate a relative arrangement and / or orientation of the terminal to the transport area and / or are characteristic thereof, are captured by an, in particular mobile, terminal of an operator, at least in sections, and are transmitted to the receiving device. Particularly preferably, the transport region can be (geometrically) measured by the, in particular mobile, terminal device and variables characteristic of a geometric extent of the transport region are preferably transmitted to the receiving device of the container treatment apparatus.In a further preferred method, the transport area size is determined automatically (by the container treatment device). This offers the advantage of a very user-friendly commissioning, in which the operator does not have to perform any measurements and / or calculations and the like.The container treatment device can preferably be executed in an adjustment mode in which the transport region size is determined, in particular optically. The setting operation preferably differs from an (intended) working operation of the container treatment device (with the highest possible production numbers or transport speeds), in which the container treatment device is executed for (normal) production or treatment of the containers. For example, the transport speed of the transport device can differ at least in the section of the transport area.In a further preferred method, image data are generated or acquired (and evaluated) by means of the image acquisition device in a setting state of the transport device in which a maximum number of containers receivable therein are recorded, in particular automatically, in the transport region in order to ascertain the transport region size.The adjustment state is understood in particular to mean a state of the transport region or of the transport device of the container treatment apparatus in which there is maximum occupancy of the transport region with containers (the containers to be treated) (i.e. in a state of the most dense packing of the containers in the transport region in which in particular the gaps between the containers are minimal).Preferably, on the basis of the captured and / or generated image data, which depict a setting state or a maximum occupancy of the transport area, the transport area size is determined, i.e. preferably that quantity and / or that number of image data points (for instance pixels) which depict a part of one of the containers in the transport area. The transport area size is preferably determined on the basis of the image data acquired and / or generated in this way by segmenting this image data. Preferably, a variable characteristic of the segment size (for example via the determination of the area and / or the number of data points) of the segment or of all those segments which map a container and / or a recognition region of the container (in particular as a transport region size) is determined.This transport area size can then preferably be used as a comparison variable with which, in a working operation of the container treatment device, the segment or segments determined therein by segmentation, which represent a container and / or a recognition area of the container or a variable derived therefrom, is compared.It is also conceivable that several different setting states (within the same container type) are possible. Thus, a slightly different arrangement of the containers, offset for example in the transport direction, can be possible. It is preferred that the image capturing device captures and / or generates image data at least once per setting state. Thus, a transport area size can be determined for each setting state and the transport area size used in a working operation of the container treatment device can be obtained by averaging over these individual transport area sizes.The use of the image capturing device offers the advantage that this image capturing device has the same configuration and / or arrangement or can be configured and / or arranged in the same way as in the working operation of the container treatment apparatus in which the image capturing device generates and / or captures the image data for determining the occupancy level variable. These two images or image data can thus be directly compared with one another.The setting operation of the transport device (or its execution) can preferably be triggered by an operator. The setting state can preferably be reached by the setting operation (and thus indirectly by triggering the setting operation). In other words, the setting state can be generated by performing the setting operation. Thus, for example, by carrying out the setting operation, it can be achieved that containers are transported into the transport region (by the transport device) until a maximum occupancy is reached. An approach and / or achievement of a maximum occupancy of the transport area can be monitored by means of the image capturing device and thereby image data captured by the transport area.Image data can thus be captured with the image capturing device at a time interval (at different capturing times) and a transport region size can be determined in each case on the basis of said image capturing device. If this increases (still) compared to previous image data of earlier capture times, it can be assumed that the maximum occupancy has not yet been reached. If, on the other hand, the transport area size remains constant, it can be assumed that a maximum occupancy has been reached.It is also conceivable for the setting operation to be carried out a plurality of times (and for the transport region to be emptied in the meantime and / or for the containers located in the transport region to be transported further) and therefore for a set state to be generated a plurality of times. It is conceivable that the transport area size is averaged and / or determined on the basis of a plurality of setting states.It is also conceivable that the transport area size is determined, for example, in the case of an empty transport area and / or independently of the occupancy of the transport area and on the basis of image data acquired and / or generated here (by means of the image acquisition device) and depicting the transport area. Thus, for example, by object recognition (which for example executes the image evaluation device or an image recognition device external to the container treatment device), a delimitation of the transport region (at least in sections) could be recognized. Preferably, on the basis of the recognized boundary of the transport region, an area and / or a number of image data points (for example pixels) which lie within the boundary is determined. These variables (or variables derived therefrom and / or variables characteristic thereof) can be used, for example, as a transport region variable.The object detection can detect and / or identify, for example, a side wall and / or a railing of the transport region and / or a lateral cover and / or an edge or a transition of the transport region to a region surrounding the latter (for example a hall floor or railing). Preferably, an (approximately lateral) boundary can be derived from this.Additionally or alternatively, it is possible that, for example via the human-machine interface and / or the receiving device of the container treatment device, at least one boundary (the full boundary or boundary) of the transport area depicted in the image data can be identified and / or specified by a user.Depending on the limitation, here, for example, an (area) size characteristic of the transport area and / or a number of image data points (which map the transport area) corresponding to the area of the transport area can be used as the transport area size.A (manual and / or automatic) geometric identification or identification of the arrangement of the mapping of the transport area in the image data offers the advantage that only image data points which map an area of the transport area are hereby used in the determination of the occupancy level size and / or the transport area size. This can ensure that image areas which are incorrectly recognized as a container (for example as a result of reflections or reflections on a metallic body, such as a railing) but which are situated outside the transport area, influence the determination of the occupancy level variable.It is also conceivable that the transport area size is determined by means of an image evaluation model of machine learning, which was trained with a plurality of image data and the associated transport area sizes.In a further preferred method, the image data are evaluated in such a way to determine the occupancy level variable that a plurality of containers, in particular adjacent containers, located in the transport area is recognized as a uniform container cluster, which does not indicate a differentiation of individual containers. This offers the advantage of a faster image data evaluation, because no individual containers or their alignment have to be distinguished. In particular, each container cluster in the transport area is assigned a (separate) segment obtained by the segmentation.In contrast to container recognition, in which each container is separately recognized and / or identified (and is labeled and / or marked as such), individual containers of a uniform container cluster are no longer distinguishable after the segmentation.In particular, the occupancy level variable is not determined using an image recognition or an evaluation of the image data, which recognizes all containers of the transport area in a form that is distinguishable from one another or individually identifiable from one another.In a further preferred method, the occupancy level variable is determined independently of a position and / or orientation of the containers located in the transport area. This also offers the advantage that a fast, robust determination of an evaluation result that is meaningful for the degree of occupancy is made possible.A segment obtained by a segmentation no longer contains, in particular, any information about a number and / or orientation of the containers located corresponding to the segment in the transport region (which are mapped by the image data points of this segment).In a further preferred method, the segmentation is a semantic segmentation. In particular, the segmentation is not an instance segmentation. This also achieves a rapid and robust determination of the occupancy level variable, which is of fundamental importance in the temporarily extremely high transport speeds of the containers that are customary in the beverage industry.In a further preferred method, a class "container" selected from a plurality of, preferably two, classes is assigned to an image data point by means of the segmentation if the image data point is assigned to a container and / or a container cluster, and / or a class "background" if the image data point is not assigned to a container and / or a container cluster.It is conceivable that such an assignment and / or assignment takes place on the basis of a detection region of a container (such as a bottle cap). Thus, a container lying on the transport area could also be detected on its bottle cap.In a further preferred method, the occupancy level variable is determined on the basis of those image data points to which the class "container" was assigned during the segmentation.In a further preferred method, the segmentation takes place on the basis of a predefined and / or predefinable color range. In particular, the segmentation can be carried out on the basis of a predetermined color filter. The color range can (at least partially and preferably completely) comprise and / or consist of those color values which the recognition range of the container and / or which the container has. The color filter is preferably selected such that it identifies those image data points which (substantially) have the color values of the recognition region and / or of the container. This advantageously allows rapidly implementable and simultaneously efficient segmentation to be realized.In a further preferred method, the segmentation is carried out on the basis of a machine learning segmentation model.The machine learning segmentation model is preferably based on an (artificial) neural network. The neural network is preferably a deep neural network (DNN), in which the parameterizable processing chain has a plurality of processing layers, and / or a so-called convolutional neural network (CNN) (dt. Convolutional Network) and / or a recurrent neural network (RNN. Recurrent neural network) and / or other DNN layers classes are formed.The machine learning segmentation model is preferably an already (fully) learned or trained segmentation model.The machine learning segmentation model has / is preferably trained with a training data set which comprises a multiplicity of image data (for feeding as input variables for the segmentation model) (imaging transport region with containers located therein) and segmented data respectively assigned thereto and obtained after a segmentation. The segmentation can be carried out as described above.The training data set can only comprise data with respect to exactly one container type and / or with respect to a predefined container treatment device. However, it is also conceivable for the training data set to comprise image data which have been acquired and / or generated in different container treatment apparatuses (of the same and / or similar construction) (and their respective segmented data). The use of image data that covers the container treatment device offers the advantage that the segmentation model trained therewith is more robust with respect to structural changes (for example, the course of the transport region) and / or brightness changes and / or environmental changes.The training data set to be generated preferably serves for (further) fine adjustment of the container recognition model or is used for this purpose.It is thus conceivable that the container recognition model has / is already trained with a general training data set, preferably independent of a specific or concrete container treatment device.In a further preferred method, at least one further transport region and preferably a plurality of transport regions are specified. Preferably, a (contiguous) transport section of the transport device is / is divided into the plurality of transport regions and is particularly preferably composed of these (in particular without overlap). The transport regions are arranged in particular along the transport path.Preferably, image data are generated for the at least one further transport area or for the respective transport areas of the plurality of transport areas by means of at least one image capturing device. In this case, a single image capturing device can capture the image data. However, it is also conceivable for a plurality of image capturing devices to gather the image data. It is also conceivable that exactly one image capturing device (one-to-one) is assigned to each transport area.Preferably, the generated image data respectively depict the containers located in the respective (at least one further) transport region.Preferably, the image evaluation device performs an evaluation, in particular a real-time evaluation, on the basis of the respective image data in each case for determining a degree of occupancy variable with respect to the respective transport area, which is characteristic of the containers located in the respective (or in the at least one further) transport area.In this case, the image evaluation device undertakes at least partial segmentation of the image data in order to determine the respective occupancy level variable.The respective occupancy level variables are preferably determined in each case in one of the above-described (preferred) methods.It is conceivable that each occupancy level variable serves as a control variable for controlling a (dedicated) drive device of the transport device. In other words, each drive device of the transport device can be assigned a transport area, depending on the (determined) occupancy level variable of which this drive device is controlled. However, it is also conceivable for a drive device to be controlled as a function of the (determined) occupancy level variables of a plurality of transport areas.Preferably, a transport section of the transport device is divided into a plurality of transport areas depending on the position of drive devices of the transport device and / or the position of areas of the transport device controlled by different drive devices.Additionally or alternatively, the subdivision of a (in particular contiguous) transport section into a plurality of transport areas (for each of which in particular an occupancy level variable is determined) can depend at least in sections on at least one geometric characteristic variable which is characteristic of an (at least in sections) geometric profile of the transport section. The geometric characteristic variable could be, for example, a variable characteristic of a curvature and / or of a width (of a section of the transport region) and / or a branching and / or a merging. Thus, with strong curvatures (small radii of curvature) and / or with comparatively large widths of the transport region in this section (compared to further sections of the transport region), a smaller subdivision into transport sections (in relation to a geometric extent along the main transport path) could be selected. Width is understood to mean, in particular, the geometric extent of the respective section perpendicular to the main transport path.In the preferred method proposed, there is in particular the focus on segmentation, in particular on the segmentation of, for example, bottle caps in filling systems, in particular in the beverage industry. The preferred approach is to use a (in particular semantic) segmentation model such as a neural network or a color filter and to regulate the performance of the container treatment device and / or of individual machines in the system on the basis of the number of detected pixels of bottle caps in relation to the total number of pixels in a specific buffer or transport section (the transport reach size).The system uses image capture devices and image processing computers to acquire data. The particularly semantic segmentation model, which may be a neural network or a color filter, is used to identify bottle caps. Advanced image analysis and segmentation techniques are employed in the associated data processing systems to identify and segment bottle caps based on pixel analyses.The process preferably includes scanning buffer and transport segments in the filling system using cameras. However, the main focus is on segmenting bottle caps on account of the number of pixels. This segmentation process is performed carefully, regardless of the position or orientation of the bottle caps. The system sums the number of detected pixels corresponding to bottle caps and compares this pixel number to the total number of pixels in the buffer or transport section. Based on this comparison, the system regulates the throughput capacity of the individual machines in the filling system.Advantages of this preferred approach, using advanced semantic segmentation models and pixel-based analysis, provide a more accurate representation of occupancy levels, which improves the performance of the machines and reduces the likelihood of transport jams. This not only results in smoother production, but also contributes to a more reliable and efficient filling process in the beverage industry.The container treatment apparatus is preferably selected from a group comprising a transport apparatus (such as a transport belt) for transporting the containers, a buffer apparatus for temporarily buffering containers, a pasteurization device (such as a tunnel pasteur), a heating apparatus for heating a preform, a forming apparatus for forming a plastic preform into a plastic bottle, a sterilization apparatus for sterilizing a container, in particular a plastic preform, a production apparatus for producing a glass bottle, a filling apparatus for filling a container with a product, an inspection device for inspecting a plastic preform or a bottle, a labeling apparatus for labeling a container, a closing apparatus for closing an, in particular filled, container, a control device, a packaging device, a direct printing device for printing a container, a collating device for collating a plurality of containers to form a collating unit or a container and the like.The container stream is preferably a (in particular continuous) stream of successively or successively following containers (on the transport path). In this case, the container stream can be guided or transported in regions and preferably within the entire container treatment apparatus (as mass flow) in a single-path or else in multiple-paths (by means of the transport device).The transport device can also be a mass transporter for the preferably multiple-track and / or disordered transport of a plurality of containers. The transport device can also be a buffer region for, preferably multiple-track and / or disordered, buffering of a plurality of containers.The containers can be transported or guided, preferably at least in sections and preferably along the entire transport region, upright (by the transport device).The transport device is preferably suitable and intended for at least sectionally guiding or transporting the plurality of containers, preferably along the entire transport region, of containers which are under dynamic pressure.The transport device is preferably suitable and determined for transporting and / or guiding (at least within the transport region) at least 1 container per hour, preferably at least 5000 (in particular to be detected) containers per hour, preferably at least 20 000, preferably at least 100 000 (in particular to be detected) containers and particularly preferably at least 140 000 (in particular to be detected) containers and does this during the working operation of the container treatment apparatus. The transport device is preferably suitable and intended for transporting and / or guiding (at least within the transport region) containers (in particular to be identified) at most 150 000 per hour and does so within the working operation of the container treatment apparatus.Preferably, the transport device is suitable and determined in a single-web transport region for transporting and / or guiding (at least within the single-web transport region) of at least 100,000 containers per hour and / or up to 150,000 containers per hour and does so within the working operation of the container treatment apparatus.The present invention is further directed to a method for at least partially automatically determining a degree of occupancy of (at least) a transport area in a container treatment device for treating containers with respect to containers located in the transport area.The container treatment apparatus has a transport device with which the containers (to be treated) can be guided and / or transported along a predefined transport path (in particular towards the transport region).The transport area is arranged along the transport path, wherein image data are generated by means of at least one image capturing device, which image containers located in the transport area and on the basis of which an image evaluation device (in particular the container treatment device) carries out an evaluation, in particular a real-time evaluation, for determining a degree of occupancy variable which is characteristic of the containers located in the transport area.According to the invention, the image evaluation device performs at least partially segmenting of the image data in order to determine the occupancy level variable.In this case, the method can have all the method steps described above in connection with the method for at least partially automatic, preferably fully automatic, control of a container treatment apparatus, in particular for determining the occupancy level variable and / or for performing the segmentation, individually or in combination with one another.The present invention is further directed to a container treatment apparatus for treating containers, having a transport device which is suitable and intended for transporting and / or guiding the containers along a predetermined transport path.The container treatment apparatus further comprises at least one image capturing device for generating image data which image containers located in the transport region.The container treatment device further comprises an image evaluation device which is suitable for determining an occupancy level variable characteristic of the occupancy level of the transport area and is intended to carry out an evaluation, in particular a real-time evaluation, on the basis of the image data.According to the invention, the image evaluation device is suitable for determining the occupancy level variable and determines to carry out at least section-wise segmentation of the image data. Furthermore, the container treatment device has a control device for at least partially automatic, preferably fully automatic, control of a container treatment device as a function of the determined occupancy level variable.In this case, the container treatment device can have all features described above in connection with the method for at least partially controlling a container treatment device and / or for at least partially automatically determining a degree of occupancy of a transport region in a container treatment device (in particular the container treatment device) solely or in combination (and vice versa) and be suitable and determined for carrying out all method steps described above in connection with the two methods (in particular also according to a preferred embodiment).Conversely, the container treatment devices mentioned above in the context of the two methods can have features of the container treatment device described here individually or in combination with one another.The present invention is furthermore directed to a method, in particular a computer-implemented method, for automatically generating a training dataset for training, in particular for retraining, a (trainable) container recognition model of machine learning (in particular a (processor-based) real-time image evaluation device) of a container treatment apparatus, in which (in particular in a working mode of the container treatment apparatus) containers can be guided along a predefined transport path, preferably in the form of a container stream, and / or transported (by means of a transport device).The container recognition model of machine learning can preferably be the above-described image evaluation model and / or segmentation model of machine learning.It is also conceivable that instead of generating a training dataset for training, in particular for retraining, the above-described container recognition model (only) the method according to the invention is directed to automatic labeling of image data (wherein a training dataset is preferably generated based on the labeled image data in a subsequent step independent or different from the method, for example by assigning generated annotation data to the respective image data).In this case, (in the working mode of the container treatment apparatus) image data can be generated by means of (at least) one image capturing device (and preferably with a plurality of image capturing devices), which image data are characteristic of a (predefined and / or predefinable and / or in particular detectable by the at least one image capturing device) transport region along the transport path (and containers located therein) and which can be fed to the container recognition model as input variables (for example for determining a degree of occupancy of the transport region) in order to carry out an evaluation, in particular a real-time evaluation, with respect to the containers located in the transport region. The image capturing device is preferably an image capturing device of the container treatment apparatus. The transport region is preferably a region within which the containers located therein are to be detected. In particular, the image data generated by means of (at least) one image capturing device (and preferably with a plurality of image capturing devices) image the containers located in the transport region.In other words, images are recorded (in the working mode of the container treatment apparatus) with the at least one image recording device, in which the transport region (and containers located therein) are recorded or are imaged in the recorded images. These are preferably evaluated in real time by the real-time image evaluation device using the container recognition model. For example, using the container recognition model as the evaluation result or as the recognition result, the real-time image evaluation device can use at least one for the number of containers recognized in the transport area and / or position (or location) of the (recognized) container (or each container recognized in the transport area) and / or orientation (or each container recognized in the transport area) and / or type of container (or each container recognized in the transport area) and / or state, the speed of the container (or each container detected in the transport area) and / or a degree of occupancy of the transport area or a section along the transport path with respect to containers located therein and / or distribution of the containers in a predetermined area within the transport area determine characteristic evaluation variables and / or provide them for output and / or transmission.The evaluation variable preferably serves as a control variable for the container treatment device, for example for (automatically) carrying out a treatment function on at least one and / or on a plurality of containers. It is also conceivable that the at least one evaluation variable is used as a control variable for controlling and / or regulating the container flow and / or a container throughput through the container treatment device and / or a transport speed and / or a container feed and / or container discharge.The image capturing device (or the plurality of image capturing devices) can be an image capturing device such as a camera (preferably black and white and / or colored), a CMOS sensor (CMOS Abk. for Complementary metal-oxide-Semiconductor), a CCD sensor, a 3D sensor, an X-ray-based image capturing device, an optical element, a thermal imaging camera, a stereo camera, a LIDAR camera and the like, and combinations thereof.The image data (generated by the image acquisition device and / or to be supplied to the container recognition model) is preferably two-dimensional (spatially resolved) image data. The image data which can be supplied or are supplied to the container recognition model for performing an evaluation preferably do not comprise any depth information (measured directly or directly by the image capturing device), that is to say in particular no (depth and / or distance) measured values measured in the direction of the recording direction of the image capturing device, which are characteristic of a distance of the image capturing device from the objects imaged in the image data. In other words, the image capturing device preferably does not generate (and / or capture) a measured value which is characteristic solely of a distance and / or a (relative) position of the image capturing device with respect to an imaged object. Such a distance and / or a position of the image capturing device relative to the objects imaged by it could be determined from image data generated by image capturing device(s) at mutually different positions (for example via stereo recordings and / or a LIDAR camera). It is thus possible to use a 2D image (recorded by the image capturing device) to calculate back to the spatial coordinates if the diameter of the detected or transported container and / or the container to be treated is known. Alternatively, a type of checkerboard with known page lengths can also be used as an approach for calculating back. Thus, a predefined reference line or reference surface (preferably lying completely in a plane) with a checkered configuration (preferably with at least two reference lines of predefined arc lengths of the lines extending at least in sections in different spatial directions) could be predefined as a reference variable for determining three-dimensional spatial coordinates for a desired 2D image data point. On the basis of the reference surface imaged in the 2D image data or from the imaged reference lines and their known or predefined geometric dimensions, a distance from the image capturing device and in particular a (3D) spatial coordinate can be determined with respect to a 2D image data point of the 2D image.Preferably, the image capturing device records a color image or a color video sequence or a color image sequence (for ascertaining and / or generating the image data). However, it is also conceivable that the recorded and / or generated and / or determined image data is a grayscale image or grayscale images. In other words, it is conceivable for achromatic image data to be supplied as input variables to the container recognition model. In this way, a larger transport region of the container treatment device can advantageously be detected or mapped and evaluated in the image data.The image capturing device (or the plurality of image capturing devices) is preferably suitable and determined for recording (static) (single) images and / or moving images or image sequences (or video sequences) or is used for this purpose.The image data (generated by the image capturing device and / or to be supplied) to the container recognition model can be image sequences and / or individual images and / or image data recorded or (substantially) simultaneously recorded or captured at a single recording time.It is conceivable that the image data (to be supplied to the container recognition model) is image data determined from more than one individual image. For example, two (mutually different) image capturing devices can each record an image (preferably substantially simultaneously). Preferably, an image merged from a plurality of images and / or image data generated from a plurality of images is supplied to the container recognition model.Preferably, pre-processed image data are supplied to the container recognition model. Preprocessing steps of the image data preferably comprise cropping, sharpening, changing the brightness (brightness). Preferably, there is no change in the color range of the image data, in particular no conversion of a color image into a grayscale image.The container recognition model of machine learning is preferably based on an (artificial) neural network. The neural network is preferably a deep neural network (DNN), in which the parameterizable processing chain has a plurality of processing layers, and / or a so-called convolutional neural network (CNN) (dt. Convolutional Network) and / or a recurrent neural network (RNN. Recurrent neural network) and / or other DNN layers classes are formed.The container recognition model of machine learning is preferably an already learned or trained container recognition model. In other words, the container recognition model to be trained or to be retrained more accurately with the training dataset to be generated is present in a state after completion of a training process. The training data set to be generated preferably serves for (further) fine adjustment of the container recognition model or is used for this purpose.It is thus conceivable that the container recognition model has / is already trained with a general training data set, preferably independent of a specific or concrete container treatment device.The method according to the invention comprises a provision and / or an evaluation of (predefined and / or predefinable) image data for generating the training dataset.The (predefined and / or predefinable) image data provided and / or to be evaluated for generating the training dataset is preferably image data generated or determined by means of (at least) one image acquisition device of a container treatment apparatus. The container treatment device is preferably that container treatment device whose container recognition model is to be trained or trained overnight with the training data set to be generated. The (at least) one image capturing device is preferably the image capturing device of that container treatment apparatus whose container recognition model is to be trained or trained overnight with the training data set to be generated. This offers the advantage that the (post-)gravating process is specifically matched to the (finely adjusted) container treatment device and, for example, specific circumstances of the specific container treatment device, such as optical properties of the image capturing device or also specific light conditions or reflection conditions or geometric conditions (for example of the container guide) in the container treatment device are (can) (can be) immediately taken into account.However, it is also conceivable that the (predetermined and / or predeterminable for generating the training dataset) have been or are generated or transmitted by a (at least sectionally and / or completely) identically constructed (but different) container treatment device.The image data (provided and / or to be evaluated and / or specified and / or predeterminable) provided for generating the training dataset are preferably recorded and / or generated and / or determined at a location (with respect to the location of generating the training dataset), preferably directly in the vicinity of or on the container treatment apparatus, the container recognition model of which is to be trained and / or trained overnight with the training data to be generated (preferably during the (current) working operation of the container treatment apparatus). The different location is therefore located in particular outside the building and / or operating site in which the life handling device is arranged and is preferably spaced apart therefrom. The image data generated or determined on or in the container treatment device (predetermined and / or predeterminable for generating the training dataset) are preferably transmitted to an external storage device (with respect to the container treatment device), which is accessed in particular via the Internet (and / or via a network, in particular a network wired and / or wireless, public and / or private network, in particular a network wired at least in sections).A sequence of image data could also be loaded into the (cloud-based) external storage device or transmitted thereto. Preferably, a training data set can be generated there on the basis of the sequence of image data. Thus, a container recognition model or the container recognition model can be trained there overnight.The method for generating a training dataset preferably comprises retrieving and / or receiving image data (the image data specified and / or predeterminable for generating a training dataset) from the external storage device.However, it is also conceivable that an image data set is received as image data by means of a user data obtained within the scope of a user input.The image data is preferably photorealistic image data. However, it is also conceivable that synthetic image data and / or augmented images (based in particular on photorealistic image data) are / are specified as image data. This offers the advantage that, for example, situations or states (for example concealed / fallen containers) which do not occur or occur only rarely in the normal working mode of the container treatment device can be simulated or cannot be adjusted and used for training.The image data provided (for generating the training dataset) and / or to be evaluated (specified and / or specified) and / or acquired with the image acquisition device preferably form a transport region along a transport path of containers to be transported in a container treatment apparatus (by a transport device), wherein at least one container and preferably a plurality of containers are preferably located in the transport region. The containers are preferably containers of the type with respect to which an evaluation is to be carried out in the machine learning container recognition model to be trained and / or to be trained.Preferably, at least a part of the (particularly preferably all) image data form the same transport region (in particular the same container treatment device). The different image data (different image recordings) preferably have different numbers and / or distributions of the containers within the transport region.According to the invention, and / or a subdivision of the transport region into at least one transport section, preferably at least two and preferably a plurality of transport sections, preferably non-overlapping one another, is provided. The individual transport sections form in each case a section of the transport region which is contiguous along the transport path. It is conceivable, for example, for the subdivision to indicate a subdivision of the transport region into at least 5, preferably at least 10, preferably at least 13 and particularly preferably at least 20 different transport sections.Preferably, such a subdivision is provided that the transport region can be composed of the plurality of transport sections. Preferably, each area of the transport area is contained in only one transport section. Preferably, the plurality of transport sections completely cover the transport region. It is also conceivable for the plurality of transport sections to cover the transport region only in sections, for example a central region of the transport region.According to the invention, image section data are determined and / or generated from the image data according to the subdivision of the transport area. In this case, (in particular for evaluating the image data to be evaluated for generating the training dataset), the image section data are fed to a training data container recognition model as input variables for evaluating the image section data with respect to the containers located in the transport section.The image section data can be determined and / or generated from the image data according to the subdivision of the transport area by cropping (or cropping) and / or subdivision and / or subdivision and / or selection of the image data in accordance with the subdivision of the transport area provided.The image section data can result from processing the image data before and / or after a selection (cropping, cropping, subdivision and / or subdivision) of the image data to determine and / or generate the image section data by sharpening and / or a change in brightness (brightness) has been made.The training data container recognition model is in particular not the machine learning container recognition model (of the container treatment device) to be trained and / or retrained with the training data set to be generated.The training data container recognition model is preferably a (trained) machine learning model. The training data container recognition model is preferably suitable and intended for mapping image data supplied as input variables, here the image section data supplied as input variables, to output variables which are characteristic of the containers located in the image data or image section data supplied as input variables.The training data container recognition model of machine learning is preferably based on an (artificial) neural network. The neural network is preferably a deep neural network (DNN), in which the parameterizable processing chain has a plurality of processing layers, and / or a so-called convolutional neural network (CNN) (dt. Convolutional Network) and / or a recurrent neural network (RNN. Recurrent neural network) and / or other DNN layers classes are formed.Preferably, by applying the training data container recognition model to the image section data supplied as input variables, at least one (computer-implemented) computer vision method is used for evaluating the image section data, in which (computer-implemented) perception and / or detection tasks are carried out, for example (computer-implemented) 2D and / or 3D object recognition methods and / or (computer-implemented) methods for (preferably semantic) segmentation and / or (computer-implemented) object classification ("image classification") and / or (computer-implemented) object localization and / or (computer-implemented) edge recognition.Preferably, the same type of computer vision method for evaluating the supplied image data can be carried out by applying the training data container recognition model than by using the container recognition model for evaluating image data supplied thereto.Preferably (with the aid of the training data container recognition model) (at least) a recognition result (as an evaluation result) is (at least) determined, preferably with respect to the respectively supplied image section data and / or the image data, on the basis of which the image section data were determined. The recognition result is preferably the same type of recognition result or evaluation variable, which is determined by means of the container recognition model.Thus, by means of the training data container recognition model or using this training data container recognition model as an evaluation result or as a recognition result, at least one position (or location) of the (recognized) container (or each container recognized in the transport section) and / or orientation (or each container recognized in the transport section) and / or type of the container (or each container recognized in the transport section) and / or state can be determined for the number of containers recognized in the transport section of the respective image section data and / or using this training data container recognition model, the speed of the container (or each container detected in the v) and / or a degree of occupancy of the transport section or a section along the transport path with respect to containers located therein and / or distribution of the containers in a predetermined area within the transport section determine characteristic evaluation variables and / or provide them for output and / or transmission.In a preferred method, all image section data belonging to the subdivision of the transport area are supplied to the training data container recognition model as input variables. Preferably, from the evaluations of the respective image section data (by means of the training data container recognition model), an evaluation of the image data (on the basis of which the image section data were determined) is carried out with respect to the containers located in the transport region.In this way, the individual evaluation results of the image section data resulting from the subdivision of the image data are reassembled and a reference to the (original) image data is established. Thus, the evaluation results can be related to those areas of the image data to which the image portion data corresponds. Thereby, evaluation results are obtained with respect to the image data (covering the transport area). Annotation data are advantageously determined from the evaluation results, which are assigned to the respective image data. In this way, a training data set is advantageously generated, which comprises the respective image data and the annotation data (determined on the basis of the image section data determined according to the subdivision (via the evaluation results thereof)).The (provided) subdivision can be applied to all image data. However, it is also conceivable for a plurality of partitions to be provided, which are applied to different image data (which can, for example, depict a different transport area or else the same transport area) for the determination and / or generation of image section data.The evaluations of the image section data instead of the image data mapping the transport region thereby offer the advantage that smaller training container recognition models can be used. This advantageously leads to a generation of a training dataset that is faster in time and less complicated (for example when the training container recognition model is provided or generated and / or with regard to a required data processing time).A smaller training container recognition model can be understood to mean, for example, a training container recognition model with parameters that are to be trained and / or retrained comparatively less. Thus, for example, a (smaller) training container recognition model can be used, which comprises (only) a set with less than 10 million, for example only 7 million, parameters to be trained and / or retrained, in comparison to a (larger) training container recognition model with a set of, for example, 100 million parameters to be trained and / or retrained.In the case of such smaller training container recognition models, the evaluation result for the individual image section data is obtained more quickly than is the case in the case of larger training container recognition models. This offers the advantage that, on the one hand, the training method can be concluded more quickly in time as a result and, on the other hand, an evaluation of supplied image data carried out in real time (in particular with respect to the working operation of the container treatment apparatus) is even made possible.Preferably, a model is used as training container recognition model, to which only image data (or image segment data) can be supplied as input variable or processed by it, which do not exceed an image size of 1000×10000 pixels, preferably 900×100 pixels, preferably 750×750 pixels and particularly preferably 640×640 pixels.The subdivision is preferably carried out in such a way that the image segment data (supplied to the training container recognition model as input variable) do not exceed an image size of 1000×10000 pixels, preferably 900×100 pixels, preferably 750×750 pixels and particularly preferably 640×640 pixels.In a preferred method, the subdivision of the transport area is predefined and / or can be predefined by an operator. Preferably, the method can comprise receiving and / or retrieving data characteristic of the (at least one) subdivision. The data characteristic of the (at least one) subdivision can preferably be input by an operator via a human-machine interface and / or, after receiving by an operator input, are stored on an (external) storage device and are retrieved from there.The subdivision is preferably changeable by an operator.In a further preferred method, the subdivision of the transport region is selected such that the transport sections produced thereby have a receiving capacity of in each case not more than 300 containers, preferably not more than 250 containers, preferably not more than 200 containers. The containers are those containers with respect to which the image data are to be evaluated. This also advantageously achieves the effect that training data container recognition models can be used to generate the training dataset, the high recognition accuracy of which is limited to supplied image data with approximately 300 or 250 or 200 containers and the recognition accuracy of which decreases with very high numbers of containers to be recognized. This achieves a high quality of the training data to be generated.In a further preferred method, the subdivision of the transport region is selected such that the maximum receiving capacity of containers of the respective transport sections is between 10 and 300, preferably between 20 and 200. This advantageously achieves the effect that, on the one hand, a rapid evaluation of the provided image data can take place and, on the other hand, the greatest possible detection accuracy is achieved.The (at least one) subdivision (provided) can be a regular subdivision. Regular is understood here to mean, in particular, a regularity with respect to a main transport path, regular subdivision. In particular, the extent or geometric extent, as seen along the main transport path, of the transport sections resulting according to the subdivision is in each case substantially the same size. The main transport path could be a medium and / or average transport path. It is also conceivable for the main transport path to be a transport path that is central in terms of geometry (wherein the geometric center relates in particular to a width of the transport region, which is in particular perpendicular to the respective transport direction).However, it is also conceivable that the subdivision is an irregular subdivision. Irregular subdivision can provide the advantage that, for example, geometric local conditions of the transport region can be taken into account, which can lead to a comparatively particularly easy and / or particularly difficult recognition situation of the containers located in the respective transport sections. Thus, for example, in the case of narrow curve sections of the transport region and / or transport sections in which the containers to be transported in these sections typically accumulate and / or even wedge / fall, a difficult recognition situation can be present. A comparatively finer subdivision is preferably selected here.In a preferred method, the subdivision depends at least in sections on at least one geometric characteristic variable which is characteristic of an (at least in sections) geometric course of the transport region. The geometric characteristic variable could be, for example, a variable characteristic of a curvature and / or of a width (of a section of the transport region) and / or a branching and / or a merging. Thus, with strong curvatures (small radii of curvature) and / or with comparatively large widths of the transport region in this section (compared to further sections of the transport region), a smaller subdivision into transport sections (in relation to a geometric extent along the main transport path) could be selected. Width is understood to mean, in particular, the geometric extent of the respective section perpendicular to the main transport path.The determination of the image section data is preferably carried out in such a way that the image data are not only divided into transport sections with respect to their arrangement and / or extent along a main transport path or the transport path, but are additionally also cut to size in a direction perpendicular to the main transport path and / or transport path, in order thereby to determine the image section data. This offers the advantage that image data points which do not image the transport region and which are not essential for an evaluation of the image data with respect to the containers located on the transport region can thereby be removed. This advantageously allows a more rapid and more precise detection of the containers to be achieved.The proportion of the image section data points which map a transport section preferably takes up more than 30%, preferably more than 40%, preferably more than 50% and particularly preferably more than 60%, of the entire image section data points of the image section data (which are fed to the training data recognition model).In a further preferred method, the division of the transport area is carried out automatically. It is conceivable, for example, that the transport area depicted in the image data is automatically recognized (via image evaluation and / or recognition methods) and a subdivision of the transport area into transport sections is (automatically) determined depending on the (geometric) course of the transport area.It is also conceivable that the subdivision of the transport region is determined iteratively. It is conceivable, for example, that a recognition accuracy is determined for a predetermined subdivision and it is determined as a function of the recognition accuracy whether a finer subdivision can / could lead to an increase in the recognition accuracy. For this purpose, the ascertained detection accuracy is preferably compared with a predefined and / or predefinable threshold value.Additionally or alternatively, it is likewise conceivable for the (determined) (geometric) profile of the transport region and / or of the main transport path in the section of the transport region to be analyzed with respect to at least one geometric characteristic variable, such as a curvature behavior, for example, and the subdivision to be determined automatically on the basis thereof.In a preferred method, the image data (provided for generating the training data set and / or specified and / or recorded with the at least one image capturing device) are present as color image data or as a color video sequence or colored image sequence. Colored image data and / or colored image section data are preferably supplied to the container recognition model and / or to the training data recognition model.The image section data supplied to the training data recognition model preferably have the same color spectrum as the corresponding image data.Specifically, in the determination of the image portion data based on the image data, a color filter is not applied. This offers the advantage that a greater detection accuracy can be achieved as a result. In contrast to alternative methods, the use of a color filter can be dispensed with in the present case because the use of a training data recognition model in conjunction with the individual processing of the image section data resulting from the subdivision can be implemented in a sufficiently efficient manner with regard to the computation effort resulting therefrom.In a further preferred method, the color values of the image section data supplied to the training container recognition model and / or of the image data supplied to the container recognition model extend substantially over the same color range as the respective or corresponding raw image data recorded by the image capturing device.In a further preferred method, the container recognition model and / or the test data container recognition model is used for, preferably semantic, segmentation of the image data or image section data (to be evaluated in each case).In the case of the, in particular semantic, segmentation, in particular each pixel of the image data or image section data or data derived therefrom is assigned a class (for classifying an object), here for example "container" or "non-container" (class annotation).The classes can be, for example, classes for classifying a container type (e.g. can | bottle | glass bottle | PET bottle and the like) and / or for classifying a respective container size (for example fill volumes, container height, diameter) and / or for classifying equipment of the container (for example body label, neck label, closure type, closure color and the like).In a further advantageous method, an occupancy level of a transport section and / or of the transport area is determined on the basis of the preferably semantic segmentation.In a preferred method, annotation data and preferably training data comprising an assignment of the generated annotation data to the corresponding image data are generated (depending on the recognition result and / or depending on the performed, preferably semantic, segmentation).Preferably, on the basis of the preferably semantic segmentation of the image section data (resulting corresponding to the subdivision of the transport area) carried out by the training data recognition model and / or using the training data recognition model, a preferably semantic segmentation (at least of the transport area) of the image data is carried out. In this case, all those image data points which do not correspond to an image section data point can be assigned the class "non-container".Annotation data (which are characteristic of the preferably semantic segmentation of the image data) is preferably generated to form the image data on the basis of the preferably semantic segmentation of the multiplicity of image section data and / or of the image data (at least of the transport region of the image data on the basis of which the image section data were ascertained).Preferably, on the basis of the performed, preferably semantic, segmentation, at least one occupancy level variable is determined which is characteristic of an occupancy level of the transport area and / or of a transport section of the image data. It is also conceivable for the occupancy level variable with respect to the transport area to be determined as a function of the respective occupancy level variables of the transport sections.The pixel-based or data point-based determination of the occupancy level offers the advantage of a very precise determination of the occupancy level.The determination of the occupancy level variable can be carried out container type-specifically.The position (and / or orientation) of a (preferably each) recognized container is preferably determined in the image data and / or the coordinates of a (preferably each) recognized container are determined (and assigned to the respective image data) in a (predefined) coordinate system (world coordinate system) of the (respective) container treatment device.In a further preferred method, annotation data and preferably training data comprising an assignment of the generated annotation data to the image data are generated as a function of the recognition result. The annotation data can comprise details about a (predefined) category, label, identification and / or identification of specific objects, localization and / or segmentation and / or video annotation (eigenpoint, polygon, bounding boxes for marking an object in the different frames). It is conceivable, for example, for the annotation data to include coordinates or position information of a recognized container in the image data. It is also conceivable for the annotation data to include or indicate a marking of the detected container, for example by means of a rectangle and / or a boundary line and / or a boundary box.It is conceivable that for each image data to be labeled, a (text) file with annotation data is generated, which is assigned to the image data. The assigned training data set preferably comprises the image data and the annotation data assigned to it.Preferably, in the manner described above, more than 100 different images or image data, preferably more than 1000 different images or image data, are labeled or annotation data is created for this purpose (and assigned to the respective image data). A training data set is preferably generated from this.The container treatment apparatus is preferably selected from a group comprising a transport apparatus (such as a transport belt) for transporting the containers, a buffer apparatus for temporarily buffering containers, a pasteurization device (such as a tunnel pasteur), a heating apparatus for heating a preform, a forming apparatus for forming a plastic preform into a plastic bottle, a sterilization apparatus for sterilizing a container, in particular a plastic preform, a production apparatus for producing a glass bottle, a filling apparatus for filling a container with a product, an inspection device for inspecting a plastic preform or a bottle, a labeling apparatus for labeling a container, a closing apparatus for closing an, in particular filled, container, a control device, a packaging device, a direct printing device for printing a container, a collating device for collating a plurality of containers to form a collating unit or a container and the like.The container stream is preferably a (in particular continuous) stream of successively or successively following containers (on the transport path). In this case, the container stream can be guided or transported in regions and preferably within the entire container treatment apparatus (as mass flow) in a single-path or else in multiple-paths (by means of the transport device).The transport device can also be a mass transporter for the preferably multiple-track and / or disordered transport of a plurality of containers. The transport device can also be a buffer region for, preferably multiple-track and / or disordered, buffering of a plurality of containers.The containers can be transported or guided, preferably at least in sections and preferably along the entire transport region, upright (by the transport device).The transport device is preferably suitable and intended for at least sectionally guiding or transporting the plurality of containers, preferably along the entire transport region, of containers which are under dynamic pressure.The transport device is preferably suitable and determined for transporting and / or guiding (at least within the transport region) at least 1 container per hour, preferably at least 5000 (in particular to be detected) containers per hour, preferably at least 20 000, preferably at least 100 000 (in particular to be detected) containers and particularly preferably at least 140 000 (in particular to be detected) containers and does this during the working operation of the container treatment apparatus. The transport device is preferably suitable and intended for transporting and / or guiding (at least within the transport region) containers (in particular to be identified) at most 150 000 per hour and does so within the working operation of the container treatment apparatus.Preferably, the transport device is suitable and determined in a single-web transport region for transporting and / or guiding (at least within the single-web transport region) of at least 100,000 containers per hour and / or up to 150,000 containers per hour and does so within the working operation of the container treatment apparatus.The present invention is further directed to a method for training, in particular for retraining, a container recognition model of machine learning, preferably a real-time image evaluation device, a container treatment device, in which containers can be guided (in a working operation of the container treatment device) along a predetermined transport path, in particular in the form of a container stream.In this case, image data can be generated (in the working operation) by means of an image capturing device, which image data are characteristic of a (depicted) transport region (preferably with respect to containers to be detected, interesting) along the transport path. The image data can be supplied to the container recognition model for performing an evaluation, in particular a real-time evaluation, with respect to the containers located in the transport region as input variables.According to the invention, the container recognition model of machine learning is trained and / or retrained with training data generated according to one of the above-described methods (in particular according to an above-described embodiment).It is therefore also proposed within the scope of the method according to the invention that labeled image data or image data provided with annotation data (as training data set) are used for training, preferably for retraining, the underlying image data of which are divided into smaller image sections to be evaluated by ascertaining image section data according to a provided subdivision. These smaller image sections (image section data) are preferably evaluated by means of a trained machine learning training container recognition model (with respect to the containers located in the transport region or the transport sections). Based on the evaluation of the individual smaller image sections, annotation data (with respect to containers imaged in the image data, for example, a degree of occupancy of a transport area) of the image data is preferably generated.The container recognition model and the container treatment device can each be configured with all the features described above in connection with the method described above individually or in combination with one another (and vice versa).The container treatment apparatus preferably records image data for generating a training dataset by means of one or more image acquisition device(s) and transmits these to an external storage device. The image data preferably (in each case) depict the transport region of the container treatment device. Image data (from the image acquisition device of the container treatment apparatus) is preferably generated at different recording times and with different distributions of containers in the transport region and determined to the external storage device.The image data stored on the external storage device are preferably retrieved (in particular by a training data generation apparatus or triggered by a training data generation apparatus). Preferably, a training data set is generated (by means of the method described above according to a preferred embodiment) on the basis of the fetched image data (by the training data generation device). In this case, the training dataset is preferably generated remotely with respect to the container treatment device.Preferably, (preferably by the container treatment apparatus) the training data set generated (by the training data generation apparatus) is called up by an external storage device.The (to be processed) data, in particular the image data recorded by means of at least one image capturing device, are preferably supplied as input variables to the container recognition model or the (artificial) neural network. The container recognition model or the artificial neural network preferably maps the input variables to output variables as a function of a parameterizable processing chain, wherein a state of the respectively recognized container, a location of a (each) recognized container, a speed of a (each) recognized container and / or a distribution and / or number of the containers in the transport area are preferably selected as the output variable.It is also conceivable for the container recognition model to determine as output variable a position variable characteristic of a (current) location / position of the (each) recognized container and / or a variable characteristic of a recognized container type and / or a degree of occupancy of a transport area and / or of a transport section and provide it (for transmission and / or output).Preferably, the container treatment device and / or the real-time evaluation device determines, as a function of the output variables of the container recognition model, a quantity characteristic of a speed of one and preferably each container recognized (in the transport area) and / or a state of one and preferably each container recognized (in the transport area) and / or a distribution and / or number of the recognized containers in the transport area.The container recognition model of machine learning or the artificial neural network is preferably (post)trained using the generated training data, wherein the parameterizable processing chain is parameterized by the training. An iterative training process is preferably selected, which is carried out until a predefined detection accuracy is achieved.The external storage device is preferably a (nonvolatile) storage device, in particular a cloud-based storage device and / or an external server (including. Storage device), wherein the storage device is accessed in particular via the Internet (and / or via a network, in particular at least in sections wired and / or wireless, public and / or private network). An external server is to be understood in particular as an external server, in particular a backend server, with respect to a container treatment apparatus and / or real-time evaluation device.The external server is, for example, a backend, in particular of a container treatment device manufacturer or of a service provider, which is configured to manage image data (in particular of a plurality of image capturing devices and / or of a plurality of container treatment devices) and / or to set container treatment devices. The functions of the backend or of the external server can be carried out on (external) server farm. The (external) server can be a distributed system.The present invention is further directed to a method for identifying containers located in a transport region (of a container treatment device) and / or for determining a degree of occupancy of a transport region of a container treatment device.According to the invention, a machine learning container recognition model, preferably trained according to the above-described method for training, in particular for retraining, a container recognition model of machine learning, is used for recognizing and / or tracking containers located in the transport area of characteristic size and / or for determining at least one (occupancy) size characteristic of a degree of occupancy of the transport area and / or of at least one transport section of the transport area.It is conceivable that a subdivision of the transport area as described above is provided. The transport area is preferably divided into transport sections according to the division and at least one occupancy variable is determined which is characteristic of an occupancy level in the respective transport section.The container recognition model is preferably suitable and determined for performing a preferably semantic segmentation (of image data) (as described above). The respective occupancy variable is preferably determined using the data point-wise or pixel-wise class assignment obtained by the preferably semantic segmentation (see, as described above, for example, classes: "container" | "non-container"). Thus, for example, to determine the respective occupancy level of the transport area or of a transport section, a ratio can be formed between the proportion of data points or pixels with the class assignment "container" and the total number of data points or pixels which map the respective transport area or transport section.The container treatment apparatus preferably has a real-time image evaluation device, in particular a processor-based image evaluation device, which is suitable and determined by means of the container recognition model of machine learning for performing an evaluation, in particular a real-time evaluation, with respect to the containers located in the transport region, for which purpose the image data can be supplied to the container recognition model (and the real-time image evaluation device) as input variables.It is therefore also proposed within the scope of the method according to the invention that a container recognition model of machine learning obtained by posttraining, as described above, is used for recognizing the containers (and deriving a recognition result) and / or determining the number and / or speed and / or distribution and / or occupancy level of the containers located in the transport area.Preferably, the container recognition model and the container treatment device can be equipped with all the features described above solely or in combination with one another (and vice versa).It is also conceivable for the container recognition model to determine as output variable a position variable characteristic of a (current) location of the (each) recognized container and / or a variable characteristic of a recognized container type and provide it (for transmission and / or output).Preferably, the container treatment device and / or the real-time evaluation device determines, as a function of the output variables of the container recognition model, a variable characteristic of a speed of one and preferably each container recognized (in the transport area) and / or a state of one and preferably each container recognized (in the transport area) and / or a distribution and / or number and / or container type of the recognized containers in the transport area and / or occupancy level of the transport area.Preferably, based on the recognition result and / or the determined number and / or distribution and / or position and / or container type, at least one recognized container is tracked. For this purpose, image data recorded at different times are used and evaluated in each case (in particular in real time). Positions of the respectively identified container are preferably determined on the basis of image data recorded at different points in time. Preferably, one or more speeds of the container are also determined on the basis of the determined positions and the recording times and / or on the basis of a transport speed effected by the transport device. Preferably, on the basis of determined positions and / or speeds of the container, an expected location range of the container is determined for at least one time in the future, preferably by using a Kalman filter.Preferably, such tracking is applied to several and preferably all containers located in the transport area.It is also conceivable that, on the basis of the characteristic (occupancy) variable determined for the occupancy level of the transport area and / or the detected containers, their positions and / or speeds and / or their determined future location areas to be expected, a virtual jam switch is provided in which a container jam resulting at a predetermined probability (preferably future) is determined or predicted. Preferably, in the case of a determined or predicted container jam, a warning message is provided for transmission and / or output to a user.It is also conceivable that different warning levels are determined depending on a degree of occupancy or a container distribution in the transport area and corresponding messages are provided for transmission and / or output to a user.The present invention is furthermore directed to a, preferably processor-based, training data generating device for automatically generating a training dataset for training, in particular for retraining, a container recognition model of machine learning of a container treatment device, in which containers can be guided along a predetermined transport path, wherein image data can be generated by means of an image capturing device, which are characteristic of a transport region along the transport path and which can be supplied to the container recognition model as input variables for carrying out an evaluation, in particular a real-time evaluation, with respect to the containers located in the transport region.The training data generation device (for generating the training dataset) is suitable and determined for evaluating image data (in particular in order to determine a recognition result).According to the invention, the training data generating device is suitable and determined for determining image section data from the image data on the basis of a provided subdivision of the transport region into at least one transport section, preferably at least two and preferably a plurality of transport sections, preferably non-overlapping one another, and for supplying the image section data as input variables to a training data container recognition model for evaluating the image section data with respect to the containers located in the transport section and for determining a recognition result.The training data generation device can be suitable, determined and / or configured for carrying out individual or a plurality of the above-described method steps of the method for training, in particular for retraining, a container recognition model of machine learning (a container treatment device).The present invention is further directed to a container treatment apparatus for treating containers, having a transport device which is suitable and intended for transporting the containers along a predetermined transport path, having an image capturing device by means of which image data can be generated which are characteristic of a transport region along the transport path.The container treatment apparatus has a real-time image evaluation device, in particular a processor-based image evaluation device, which is suitable and determined by means of a container recognition model of machine learning for performing an evaluation, in particular a real-time evaluation, with respect to the containers located in the transport region, for which purpose the image data can be supplied to the container recognition model (and the real-time image evaluation device) as input variables.The container recognition model of machine learning is preferably a container recognition model which has been trained and / or retrained with a training data set generated according to one of the above-described methods.The containers are preferably treated as a function of the recognition result of the container recognition model or (at least) an output variable of the container recognition model or as a function of the containers recognized by the container recognition model.In this case, the container treatment device can have all features described above in connection with a container treatment device alone or in combination (and vice versa) and be suitable and intended for carrying out all method steps described above in connection with the method for identifying containers located in a transport region of a container treatment device.The container recognition model can have one or more features described above individually or in combination.The present invention is furthermore directed to a computer program or computer program product comprising program means, in particular a program code which represents or encodes at least some of the and preferably all method steps of the respective (above-described) methods according to the invention (for generating a training dataset and / or a method for training a container recognition model of a container treatment apparatus) and preferably one of the (above-described) preferred embodiments and is designed for execution by a processor device.The present invention is further directed to a data memory on which at least one embodiment of the computer program according to the invention or a preferred embodiment of the computer program is stored.The present invention has been described with reference to containers to be identified. The present invention can also be transferred to objects to be detected generally in machine image processing (based on image evaluation models of machine learning) for image detection in the beverage and / or pharmacy sector, in particular in container treatment devices in the beverage and / or pharmacy sector. The applicant reserves the right to also claim objects directed to this, in particular to the method for generating a training dataset and the training data generation device.Further advantages and embodiments are evident from the attached drawings: FIG. 1 shows a schematic illustration of an image recording from a container treatment device; FIG. 2 is an illustration of a preferred method for generating a training dataset; FIGS. 3 a, 4 a, 5 a show different input images (to be evaluated by a container recognition model); and FIGS. 3 b, 4 b, 5 b show the respectively corresponding evaluation data obtained by the evaluation of the input images from FIGS. 3 a, 4 a, 5 a; and FIG. 6 shows a roughly schematic structure of a container treatment device according to a preferred embodiment of the present invention.FIG. 1 shows a schematic representation of an image recording BD 1 from a container treatment apparatus 1. the container treatment apparatus 1 has a transport device 2 (here a mass transporter in which the containers 10 (here cans) are transported in a multi-row and disordered manner). The reference numerals 16 and 18 identify lateral boundaries or guides of the transport device 2, which can define, for example, the region of the transport device, for example a transport beltIt can be seen that the transport region of the transport device 2 shown is divided into different transport sections which are identified in FIG. 1 by the reference symbols B 1, B 2, B 3, B 4,..., B 9, B 10.An evaluation of the image recording using a container recognition model of machine learning can result in the respective numbers of containers within the respective transport sections B 1, B 2,..., B 10. Alternatively or on the basis of the numbers of containers, a respective occupancy level of the respective transport section can be specified and / or determined (which corresponds to the ratio between the containers currently being transported in the transport section and the maximum occupancy number).In the transport section B 1, there are currently 54 containers being transported in the image data BD 1. For a maximum occupancy count (assumed here) of 119 containers, this corresponds to an occupancy level of 45.38%.At the recording time of the image data BD1, no containers 10 are currently being transported in the transport sections B2-B5. In these sections, the occupancy level is therefore 0%.The transport sections B 6 and B 7 are only partially, but not completely, occupied by containers 10.In the transport sections B 9 and B 10, in the image data BD 1, in each case just the maximum occupancy number of containers that can be accommodated in the respective transport sections is transported. In the transport sections B 9 and B 10, a (instantaneous) occupancy level of 100% results here.FIG. 1 also shows an illustration of a preferred method for generating a training dataset. In this case, the transport region depicted in the recorded image data BD 1 is broken down into individual transport sections B 1,..., B 10 and image data (of the respective entire image BD 1) cut to the individual transport sections is supplied to an image evaluation model (for example a segmentation model) and / or a container recognition model as input image data for evaluation with respect to the containers 10 located in the transport sections.The applicant has recognized in complex series of experiments that, if the image evaluation model or the recognition model is applied to a complete image BD 1, the recognition accuracy is low, and after the cropping (illustrated here by the partial region B 3, for example), all objects (here containers or doses) are recognized.This therefore means that an image evaluation model or a container recognition model with preprocessing such as (trimming, sharpness and / or brightness) is preferably used as the author.FIGS. 2 and 4 each show different input image data (input images) E 1, E 2, which in turn depict a container treatment apparatus 1 with a transport device 2 and containers 10 (here doses) transported by it and which are evaluated by means of an, in particular processor-based, image evaluation device (preferably with an image evaluation model or a recognition model of machine learning).The reference numerals 16 and 18 again identify a lateral delimitation of that receiving region 14 of the transport device 2 in which containers 10 can be received for transporting the latter.The image evaluation device (for example the recognition model of machine learning) preferably carries out (semantic) segmentation here. In the (semantic) segmentation, a class is assigned to each pixel of an (input) image or to each data point of the input image data. Here, the class "container" (or "can") is used here. All image data or pixels which belong to a (detected) container (or can) are assigned the class "container". These image data points or pixels are shown white in the evaluation images A 1, A 2 in FIGS. 3 and 5 and are denoted by the reference symbol P 10. The class "container" (or corresponding to the class "can") is to be understood here in particular as a class "image points (belonging to any container (of the container mass flow)", because individual containers or individual doses cannot be assigned to the classified points. Even if a plurality of "container pixels", i.e. data points of the class "container", are recognized, the evaluation result does not include, in particular, the information as to whether it is one or more containers (or whether it is one or more doses) or whether, for example, two adjacent data points of the class "container" belong to the same container or map (adjacent) regions of the same container.All other image data points which do not belong to objects identified as containers (or as doses) are assigned a pixel value which is different from the class "container" (or "can"), that is to say, for example, a pixel data point colored black, a "background pixel value" PH.In FIGS. 3 and 5, all image data points that the machine learning recognition model has recognized as not belonging to a container (or can) are mapped as black pixels.In these figures, a binary representation of white pixels (container) and black pixels (non-container) is obtained.Preferably, an occupancy level (of a predefined transport area) of a transport device can be determined in a quick manner via such an evaluation, for example via the determination of the white image portions P 10 of the respective evaluation images A 1, A 2.FIG. 6 shows a roughly schematic structure of a container treatment device 1 for treating containers 10 according to a preferred embodiment of the present invention. The reference numeral 2 identifies a transport device which transports the containers 10 to a treatment device 12 for the treatment of the containers 10.The reference numeral 4 designates an image capturing device, here a camera, which - here from above - captures a transport region designated by the reference numeral 20. In particular, image data are generated which depict the containers 10 located in the transport region 20.The captured image data (possibly after preprocessing of the latter) are transmitted to an image evaluation device 6. The image evaluation device 6 performs an evaluation on the basis of the image data in order to determine an occupancy level variable which is characteristic of the occupancy level of the transport area 20.The image evaluation device 6 determines the occupancy level variable based on an at least sectional segmentation of the image data. Preferably, a control variable for controlling the container treatment device 1 is determined on the basis of the occupancy level variable.The reference numeral 8 identifies a control device which preferably comprises the container treatment apparatus 1 and which serves for controlling the container treatment apparatus as a function of the control variable, for example for (automatically) carrying out a treatment function on at least one and / or on a plurality of containers. It is also conceivable that the at least one control variable is used for controlling and / or regulating the container flow and / or a container throughput through the container treatment device and / or a transport speed and / or a container feed and / or container discharge.The applicant reserves the right to claim all the features disclosed in the application documents as essential to the invention, provided they are novel, individually or in combination, compared with the prior art. It is also pointed out that features have also been described in the individual figures, which features may be advantageous per se. The skilled person directly recognizes that a specific feature described in a figure can also be advantageous without the adoption of further features from this figure. Furthermore, the skilled person recognizes that advantages can also result from a combination of a plurality of features shown in individual figures or in different figures.List of reference characters1 Container treatment apparatus 2 Transport device 4 Image acquisition device 6 Image evaluation device 8 Control device 10 Container 12 Container treatment device 20 Transport region B 1, B 2,..., B 10 Transport sections BD 1, BD 2, BD 3 Image data TB 1 Partial image E 1, E 2, E 3 Input images PH Background pixel P 10 Pixels to which the class "container" is assignedReferences included in the specificationThis list of documents cited by the applicant has been produced in an automated manner and is only included for the better information of the reader. The list is not part of the German patent application or utility model application. The DPMA does not take any adhesion for any faults or omissions.Patent Literature citedEP 2 132 129 B1

[0015] EP 300 523 1 B1

[0016]

Claims

Method for at least partially automatic, preferably fully automatic, control of a container treatment apparatus (1) for treating containers (10), which has a transport device (2) for guiding containers (10) along a predetermined transport path, as a function of a degree of occupancy of at least one transport region (20) along the transport path, wherein image data (E1, E2) are generated by means of at least one image capture device (4), said image data imaging containers (10) located in the transport region (20) and on the basis of which an image evaluation device (6) performs an evaluation, in particular a real-time evaluation, for determining a degree of occupancy of the transport region (20), characterized in that the image evaluation device (6) performs an at least partial segmentation of the image data (E1, E2) for determining the degree of occupancy.Method according to Claim 1, characterized in that the occupancy level variable is determined on the basis of at least one segment obtained by the segmentation and preferably at least one plurality of segments obtained by the segmentation.Method according to one of the preceding claims, characterized in that the occupancy level variable is determined as a function of a transport area variable which is characteristic of an occupancy area and / or of a maximum occupancy number of the transport area.Method according to the preceding claim, characterized in that the transport region size is determined automatically.Method according to one of the two preceding claims, characterized in that, in order to determine the transport region size, image data are generated (and evaluated) by means of the image acquisition device (4) in a setting state of the transport device (2), which setting state can preferably be reached by an operator by triggering a setting operation of the transport device (2), in which setting state a maximum number of containers (10) which can be accommodated therein are recorded, in particular automatically.Method according to one of the preceding claims, characterized in that, in order to determine the occupancy level variable, the image data are evaluated in such a way that a plurality of containers (10) which are located in the transport region (20) and adjoin one another is recognized as a uniform container cluster which does not indicate a differentiation between individual containers (10).Method according to one of the preceding claims, characterized in that the occupancy level variable is determined independently of a position and / or orientation of the containers (10) located in the transport region (20).Method according to one of the preceding claims, characterized in that the segmentation is a semantic segmentation.Method according to one of the preceding claims, characterized in that by means of the segmentation an image data point is assigned a class "container" (P10) selected from a plurality, preferably two, classes if the image data point is assigned to a container and / or a container cluster, and a class "background" (PH) is assigned if the image data point is not assigned to a container and / or a container cluster.Method according to the preceding claim, characterized in that the occupancy level variable is determined on the basis of those image data points (P10) to which the class "container" (P10) was assigned during the segmentation.Method according to one of the preceding claims, characterized in that the segmentation is carried out on the basis of a predefined and / or predefinable colour range.Method according to one of the preceding claims, characterized in that the segmentation is carried out on the basis of a machine learning segmentation model.Method according to one of the preceding claims, characterized in that at least one further transport region is predetermined (B7) and image data are generated by means of at least one image capturing device (4) which image containers (10) located in the at least one further transport region (20) and on the basis of which an image evaluation device (6) carries out an evaluation, in particular a real-time evaluation, for determining a degree of occupancy variable which is characteristic of the containers (10) located in the at least one further transport region (20), wherein the image evaluation device (6) carries out a segmentation of the image data at least in sections in order to determine the degree of occupancy variable.Method for at least partially automatic determination of a degree of occupancy of a transport area in a container treatment apparatus (1) for treating containers (10) with respect to containers (10) located in the transport area (20), wherein the container treatment apparatus (1) has a transport device (2) with which the containers (10) can be guided along a predetermined transport path, wherein the transport area (20) is arranged along the transport path, wherein image data (E1, E2) are generated by means of at least one image acquisition device (4) which image containers (10) located in the transport area (20) and on the basis of which an image evaluation device (6) carries out an evaluation, in particular a real-time evaluation, for determination of a degree of occupancy variable which is characteristic for the containers (10) located in the transport area (20), characterized in that, the image evaluation device (6) carries out at least a segment of the image data (E1, E2) in order to determine the occupancy level variable.Container treatment apparatus (1) for treating containers (10), having a transport device (2) which is suitable and intended for transporting the containers (10) along a predetermined transport path, having at least one image capturing device (4) for generating image data (E1, E2) which image containers (10) located in the transport region (20), and having an image evaluation device (6) which is suitable and intended for determining an occupancy level variable which is characteristic of the occupancy level of the transport region (20), for carrying out an evaluation, in particular a real-time evaluation, on the basis of the image data (E1, E2), characterized in that the image evaluation device (6) is suitable and intended for determining the occupancy level variable for carrying out a segmentation of the image data (E1, E2) at least in sections, and in that the container treatment device (1) has a control device (8) for at least partially automatic, preferably fully automatic, control of a container treatment device (1) as a function of the determined occupancy level variable.

Images