Device and method
By calculating image division sizes based on object size and center of gravity, the method reduces object cutting and enhances detection accuracy in object detection systems.
Patent Information
- Application Number
- PCT/JP2025/016726
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-05-07
- Publication Date
- 2026-11-12
AI Technical Summary
Existing image division methods in object detection systems increase the probability of cutting objects, reducing detection accuracy, while minimizing divisions to maintain pixel count makes it difficult to detect small objects.
An image division method that calculates a division size based on object size and center of gravity, using optimization techniques to minimize the probability of object cutting and enhance detection accuracy.
Improves object detection accuracy by reducing the likelihood of objects being cut during image division, allowing for precise detection of small objects.
Smart Images

Figure JP2025016726_12112026_PF_FP_ABST
Abstract
Description
Apparatus and method
[0001] This disclosure relates to apparatus and method.
[0002] Patent Document 1 describes an information processing device for detecting objects from an image. In this information processing device, an image division unit divides an input image to generate multiple divided images, and an object detection unit detects objects from each divided image. For example, the image division unit divides a 720 x 480 pixel input image into six regions and generates six 240 x 240 pixel divided images.
[0003] Japanese Patent Publication No. 2024-2327
[0004] In the technology described in Patent Document 1, the probability of cutting an object when dividing the image increases, which may reduce the accuracy of object detection. If the number of divisions of the image is reduced in an attempt to reduce the size of object cutting, the number of pixels in the resulting image remains high, which may make it difficult to detect small objects in the resulting image.
[0005] This disclosure aims to provide an apparatus and method capable of improving the accuracy of object detection.
[0006] The apparatus of this disclosure comprises an image acquisition unit that acquires an image of an object; a size acquisition unit that acquires the object size of the object based on the image; a calculation unit that calculates a division size for dividing the image based on the object size; and a division unit that divides the image based on the division size.
[0007] According to the above device, the division size is calculated based on the object size of the object in the image, making it possible to calculate a division size appropriate to the object size. This makes it possible to reduce the probability of an object being cut without excessively reducing the number of divisions, and consequently improve the accuracy of object detection in the image.
[0008] According to this disclosure, it will be possible to improve the accuracy of object detection.
[0009] Figure 1 is a conceptual diagram showing the method for calculating the division size according to the present disclosure. Figure 2 is a conceptual diagram of an object detection system including a detection device according to one embodiment of the present disclosure. Figure 3 is a flowchart of the detection method performed by the detection device shown in Figure 2. Figure 4 is a diagram showing an example of an image to which annotation data has been added. Figure 5 is a diagram showing an example of the process for obtaining representative points from the distribution of centroids. Figure 6 is a diagram showing an image divided by a virtual division size. Figure 7 is a diagram showing the state in which an object is cut. Figure 8 is a diagram showing an image divided by a virtual division size. Figure 9 is a diagram showing an example of the hardware configuration of a detection device according to one embodiment of the present disclosure.
[0010] Embodiments of this disclosure will be described in detail below with reference to the drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant descriptions are omitted.
[0011] The object detection system according to this embodiment is a system that detects objects in an image. The object detection system divides the image containing the object and detects the object from the divided image.
[0012] An overview of the object detection method in this disclosure will be described with reference to Figures 1 and 2. Figure 1 is a conceptual diagram showing the method for calculating the division size in this disclosure. Figure 2 is a conceptual diagram of an object detection system 1 including a detection device 10 according to one embodiment of this disclosure. As shown in Figure 1, the detection device 10 acquires an image to which annotation data about an object has been added. Based on the acquired image, the detection device 10 calculates statistical values of the object size and extracts representative points of the object. Details of the annotation data, object size, and representative points will be described later. Based on the statistical values of the object size and the representative points, the detection device 10 performs an optimization calculation of the division size when dividing the image. Based on the obtained division sizes, the detection device 10 divides the image and detects objects from the divided images.
[0013] The object detection system 1 shown in Figure 2 includes a drone D and a detection device 10. The detection device 10 is connected to the drone D via a network for communication.
[0014] Drone D uses a high-resolution camera to photograph the ground from a height of several meters or more, capturing a wide range of objects OB at once. Aerial photography by Drone D can be used for purposes such as pesticide spraying, and can reduce the time required for photography. Drone D outputs images of objects OB to detection device 10. The acquisition of these images may be performed using means other than Drone D. Examples of means other than Drone D include aircraft, helicopters, and satellites.
[0015] Drone D may acquire multiple images taken within the same shooting range. Increasing the number of images in this way can improve the accuracy of object detection. Each of the multiple images may contain objects of multiple classes. A class refers to the classification of an object. The objects in an image may belong to only one class.
[0016] The detection device 10 includes an image acquisition unit 11, a centroid acquisition unit 12, a size acquisition unit 13, a calculation unit 14, a division unit 15, and a detection unit 16.
[0017] The image acquisition unit 11 acquires images taken by the drone D. The image acquisition unit 11 may also add annotation data about the object to the image. The annotation data may include, for example, the object's class and a rectangular bounding box surrounding the object. The bounding box may be represented, for example, by the coordinates of the four corners of a rectangular frame in the XY coordinate system.
[0018] The center of gravity acquisition unit 12 acquires the center of gravity of an object in the image. The center of gravity is represented as coordinates in the XY coordinate system. For example, the center of gravity acquisition unit 12 considers a bounding box as an object and acquires the center of gravity of the bounding box as the center of gravity of the object.
[0019] The size acquisition unit 13 acquires the object size of an object appearing in an image. The object size is information indicating the size of the object. For example, the size acquisition unit 13 regards a bounding box as the object, and acquires the size of the bounding box as the object size. Examples of the size of the bounding box include the vertical length or horizontal length of the bounding box, and the area of the bounding box. The size of the bounding box may be half the vertical length or half the horizontal length.
[0020] The calculation unit 14 calculates a division size for dividing an image. The division size is information indicating the size of each region when the image is divided into a plurality of regions. For example, the division size may be the vertical length or horizontal length of each region, or may be the area of each region.
[0021] The division unit 15 divides the image into a plurality of regions to generate a plurality of divided images.
[0022] The detection unit 16 detects an object appearing in the divided image. The detection unit 16 detects the object using a known object detection algorithm.
[0023] The detection method performed by the detection device 10 will be described with reference to FIG. 3. FIG. 3 is a flowchart of the detection method performed by the detection device 10 shown in FIG. 2.
[0024] In step S1, the image acquisition unit 11 acquires an image in which an object appears from the drone D. The image acquisition unit 11 may acquire one image, or may acquire a plurality of images.
[0025] The image acquisition unit 11 may receive input of annotation data from a user, and attach the annotation data to the image. Alternatively, the image acquisition unit 11 may automatically generate the annotation data. Alternatively, annotation data relating to some objects may be input by the user, and annotation data relating to other objects may be generated by the image acquisition unit 11. Thereby, a rectangular bounding box enclosing the object and the class of the object are set in the image.
[0026] FIG. 4 is a diagram illustrating an example of an image provided with annotation data. In the example of FIG. 4, the image acquisition unit 11 acquires three images in which a dog, a bicycle, and a truck are captured respectively. The positions of objects such as the dog in each image may be different. For each image, a bounding box BB surrounding the object and a class CL of the object are set. In these three images, the number of object classes is three. The number of objects (dogs, bicycles, or trucks) in each class is one per image, for a total of three.
[0027] In step S2, the centroid acquisition unit 12 acquires the centroid of the bounding box as the centroid of the object based on the bounding box set for each of the plurality of images. For example, the centroid acquisition unit 12 calculates the centroid of the bounding box based on the coordinates of the four corners of the bounding box.
[0028] In step S3, the centroid acquisition unit 12 acquires a representative point of the centroid of the object for each object class. For example, when a plurality of objects belong to the same class, the centroid acquisition unit 12 calculates a representative point representing the respective centroids of the plurality of objects. The centroid acquisition unit 12 may calculate the average value of the respective centroids of the plurality of objects as the representative point.
[0029] The centroid acquisition unit 12 may calculate the representative point based on the distribution of a plurality of centroids (coordinate group of centroids). FIG. 5 is a diagram illustrating an example of a flow of calculating representative points from the distribution of centroids. On the left side of FIG. 5, the respective centroids P of seven objects belonging to the same class are plotted. The centroid acquisition unit 12 acquires the distribution of the centroids P of the objects as shown on the right side of FIG. 5, using, for example, kernel density estimation. The centroid acquisition unit 12 extracts representative points from the obtained distribution. The centroid acquisition unit 12 may extract representative points in a number specified by a user or a preset number. The centroid acquisition unit 12 may set the centroid of a region exceeding a threshold in the distribution of the centroids P as the representative point. In the example of FIG. 5, the centroid acquisition unit 12 may extract the coordinates of the vertices of four peaks Pe formed in the distribution as the representative points.
[0030] As described above, the centroid acquisition unit 12 may acquire multiple representative points for each class of object. Since objects belonging to the same class may not appear in the same position across multiple images, it is effective to acquire representative points while considering the distribution of the centroid in this way.
[0031] The center of gravity acquisition unit 12 acquires the center of gravity of an object belonging to a class as the representative point of that class, if that object belongs to a class.
[0032] In step S4, the size acquisition unit 13 acquires the size of the bounding box as the object size based on the bounding box set for each of the multiple images. The size acquisition unit 13 calculates statistical values of the bounding box size for each class of object. Examples of statistical values include the mean and the median.
[0033] For example, if multiple objects belong to the same class, the size acquisition unit 13 calculates a statistical value for the size of the bounding box corresponding to each of the multiple objects. The size acquisition unit 13 may calculate the average of the vertical or horizontal lengths of the bounding boxes as the statistical value, or it may calculate the average of the area of the bounding boxes as the statistical value. The size acquisition unit 13 may also calculate the average of half of the vertical or horizontal lengths of the bounding boxes as the statistical value. In this way, a statistical value for the size of each object is calculated for each class.
[0034] The size acquisition unit 13 acquires the size of the bounding box corresponding to the object as a statistical value for the class, if only one object belongs to that class.
[0035] In step S5, the calculation unit 14 calculates the division size for dividing the image based on the object size. For example, the calculation unit 14 calculates the division size based on the object size and representative points. The object size and representative points are used, for example, in the objective function and constraints described later.
[0036] The calculation unit 14 calculates the division size based on the virtual division size and the cutting size of the object when the image is divided by the virtual division size. The virtual division size is the division size set when the image is hypothetically divided. The cutting size indicates the size by which the object extends beyond the divided image when the image is divided by the virtual division size (the size by which the object extends beyond the divided image). For example, the cutting size indicates the size of the cut portion that is cut and separated by the grid within the area enclosed by the bounding box. For example, if the object is divided by the grid into two parts, one that includes the center of gravity of the bounding box and one that does not, the cut portion corresponds to the part that does not include the center of gravity of the bounding box. For example, if the object is divided by the grid into two parts, the cut portion corresponds to the smaller part. The grid is a grid-like division line set by the virtual division size. The cutting size may be the vertical length or horizontal length of the cut portion, or it may be the area of the cut portion.
[0037] The calculation unit 14 calculates the virtual division size as the division size when the objective function relating the sum of the virtual division size and the cutting size is minimized. The details of the calculation method will be explained with reference to Figure 6. Figure 6 shows an image divided along a grid Gr set by the virtual division size S. In the example in Figure 6, the division size is calculated in one dimension (the horizontal division width is calculated). In Figure 6, the bounding box BB of a certain class of object is shown, and the grid Gr around the representative point RP of that class is shown. The illustration of other grids Gr is omitted. N in Figure 6 is a natural number.
[0038] The calculation unit 14 sets each parameter as follows: X: Multiplier for the virtual division size (decision variable) S: Virtual division size c: Number of the object class i: Representative point number for each object class M c,i : A multiplier (auxiliary variable) used to represent the coordinates of the grid closest to the representative point: G c,i : Coordinates of the representative point W c :Statistical value of object size Z c,i: Penalty term (auxiliary variable) representing the cut size of an object β: Parameter (set value) for adjusting the weight of the penalty term A: Image size (set value) B: Divisor of the virtual division size (set value that satisfies A≡0 (mod B))
[0039] S is calculated as X×B. B is a numerical value that evenly divides the image size A, and is, for example, 32 pixels or 64 pixels. B is appropriately set according to the requirements of object detection by the detection unit 16 and the like. The object class number c is a number for identifying an object class, and in the example of FIG. 4, identifies a dog, a bicycle, and a truck. The representative point number i identifies the representative point calculated in step S3. M c,i is a multiplier for expressing the coordinates of the grid Gr close to the representative point RP. In the example of FIG. 6, the coordinates of the grid Gr closest to the representative point RP are (N+1)×S, and M c,i is (N+1).
[0040] G c,i is the coordinate of the representative point corresponding to the representative point number i of the class represented by the class number c, and is calculated in step S3. W c is a statistic of the object size of the object belonging to the class represented by the class number c, and is calculated in step S4. In FIG. 6, the object size is half the horizontal length of the bounding box BB. W c may be the object size.
[0041] In the example of FIG. 6, Z c,i is the length in the horizontal direction of the cut portion CP that is cut and separated by the grid Gr represented by (N+1)×S, among the regions enclosed by the bounding box BB. The bounding box BB is divided into two parts by the grid Gr represented by (N+1)×S. The cut portion CP is the smaller one of the two divided parts (the part on the right side of the grid Gr represented by (N+1)×S). The cut portion CP does not include the representative point RP. The calculation unit 14 calculates Z c,i based on the virtual division size, the object size, and the center of gravity (or representative point). In the example of FIG. 6, the calculation unit 14 calculates Z c,i as W c - |(N+1)×S - Gc,i It is calculated by |. Z c,i It is treated as an auxiliary variable.
[0042] The calculation unit 14 uses the above parameters to generate an objective function relating to the sum of the virtual division size and the cutting size, as well as constraints. For example, the objective function is expressed by equation (1) below, and the constraints are expressed by equations (2) to (6) below.
[0043]
[0044] Equation (1) means minimizing the sum of the virtual division size and the cutting size (separation penalty). Equation (2) means maintaining the separation between the grid closest to the object's centroid (representative point) and the object's centroid. Equation (3) means making the virtual division size a multiple of the set value B. Equation (4) means that the grid coordinates fit within the image size A. Equation (5) means that the parameter on the left side is a natural number. Equation (6) means that the parameter on the left side is a real number.
[0045] The calculation unit 14 uses mathematical optimization to determine the value that minimizes the division size for each object class, while maintaining the distance from the representative point to the grid equal to the object size. Specifically, the calculation unit 14 uses an optimization method to find the values of X, M, and Z when the objective function expressed in equation (1) is minimized under the constraints of equations (2) to (6). Examples of optimization methods include the branch cut method and metaheuristics. The calculation unit 14 calculates a virtual division size S obtained by multiplying the calculated X by B as the division size. This division size is the division width in the horizontal direction. The calculation unit 14 may also calculate the division width in the vertical direction in the same way as the division width in the horizontal direction.
[0046] In step S6, the division unit 15 divides the image into equal parts based on the division size, generating multiple divided images. The vertical and horizontal lengths of each divided image correspond to the division size calculated in step S5.
[0047] In step S7, the detection unit 16 detects objects in the divided images based on the divided images. The detection unit 16 may classify the multiple divided images based on the detection results.
[0048] In the detection device 10 and detection method described above, the division size is calculated based on the object size of the object in the image, thereby enabling the calculation of a division size appropriate to the object size. This makes it possible to reduce the probability of an object being cut without excessively reducing the number of divisions, and consequently improve the detection accuracy of objects in the image.
[0049] The above functions and effects are supplemented with reference to Figure 7. Figure 7 shows a state in which object OB is cut off. When training an object detection model without reducing the resolution of high-resolution images captured by drone D, a certain amount of GPU memory is required. On the other hand, when reducing the resolution of the image, the number of pixels of small objects decreases and they become distorted, making them difficult to detect. Therefore, a method of dividing the image into a predetermined number of parts before performing object detection can be considered. However, if the image is divided equally into a predetermined number of parts (16 in the example of Figure 7), object OB may be cut off by the grid Gr, which may reduce the detection accuracy. With the detection device 10, an appropriate division size for the object size is calculated, so the probability of object OB being cut off can be reduced.
[0050] The calculation unit 14 calculates the division size based on the object size and center of gravity. With this configuration, it is possible to calculate a division size that is appropriate for the object size and the position of the object's center of gravity. This makes it possible to further reduce the probability of the object being cut.
[0051] The size acquisition unit 13 acquires the object size based on the bounding box, and the center of gravity acquisition unit 12 acquires the center of gravity based on the bounding box. With this configuration, the object size and center of gravity can be easily acquired.
[0052] The center of gravity acquisition unit 12 acquires a representative point of the center of gravity of each object class, and the calculation unit 14 calculates the division size based on the object size and the representative point. With this configuration, it is possible to calculate a division size appropriate to the object size and the representative point. This makes it possible to further reduce the probability of the object being cut.
[0053] The center of gravity acquisition unit 12 acquires a representative point based on the center of gravity of each of the multiple objects belonging to the same class. With this configuration, when multiple objects belong to the same class, the division size can be efficiently calculated by using the representative point.
[0054] The calculation unit 14 calculates the division size based on a virtual division size and a cutting size that indicates the extent to which an object will extend beyond the divided image when the image is divided by the virtual division size. With this configuration, a more suitable division size can be calculated by taking the cutting size into consideration.
[0055] The calculation unit 14 calculates the virtual division size as the division size when the objective function relating the sum of the virtual division size and the cutting size is minimized. With this configuration, it is possible to calculate a division size that reduces the cutting size of the object while reducing the overall division size.
[0056] The calculation unit 14 calculates the virtual partition size when the objective function is minimized using an optimization method. With this configuration, the virtual partition size when the objective function is minimized can be easily calculated.
[0057] The calculation unit 14 calculates the cutting size based on the virtual division size, object size, and center of gravity. This configuration allows for easy calculation of the cutting size.
[0058] The detection unit 16 detects objects in the segmented images, which are divided from the original image. This configuration allows for high-precision detection of objects in the image.
[0059] Although one embodiment of the present disclosure has been described in detail above, the present disclosure is not limited to the above embodiment. The apparatus and method relating to the present disclosure can be modified in various ways without departing from the spirit thereof.
[0060] In the above embodiment, the calculation unit 14 calculated the division size in the horizontal direction and the division size in the vertical direction by solving separate optimization problems. However, the calculation unit 14 may also calculate the division size in the horizontal direction and the division size in the vertical direction by solving a single optimization problem. The details of this calculation method will be explained with reference to Figure 8. Figure 8 shows a virtual division size S. x and virtual partition size S y This figure shows an image divided along a grid Gr set by a virtual division size S. x This is the virtual division size in the horizontal direction, and the virtual division size S y This is the virtual division size in the vertical direction. N in Figure 8 x and N y It is a natural number.
[0061] The calculation unit 14 sets each parameter as follows: X: Multiplier for the virtual division size in the horizontal direction (decision variable) Y: Multiplier for the virtual division size in the vertical direction (decision variable) S x : Virtual division size in the horizontal direction S y : Virtual division size in the vertical direction c: Object class number i: Representative point number for each object class M x,c,i : A multiplier (auxiliary variable) used to represent the X coordinate of the grid closest to the representative point: M y,c,i : A multiplier (auxiliary variable) used to represent the Y coordinate of the grid closest to the representative point: G x,c,i : X-coordinate of the representative point G y,c,i : Y coordinate of the representative point W x,c : A statistical measure of object size in the horizontal direction W y,c : A statistical measure of object size in the vertical direction Z x,c,i : A penalty term (auxiliary variable) representing the cross-section size of the object in the lateral direction: Z y,c,i : Penalty term (auxiliary variable) representing the cutting size of the object in the vertical direction β: Parameter (set value) that adjusts the weight of the penalty term A x Image size in the horizontal direction (set value) A y : Image size in the vertical direction (set value) B: Multiplier of the virtual division size (A x ≡ 0 (mod B) and A y(Settings that satisfy ≡0 (mod B))
[0062] The calculation unit 14 uses the above parameters to generate an objective function relating to the sum of the virtual division size and cutting size, as well as constraint conditions. The objective function is expressed by equation (7) below, and the constraint conditions are expressed by equations (8) to (12) below. Equations (7) to (12) correspond to equations (1) to (6), respectively.
[0063]
[0064] The calculation unit 14 uses an optimization method to find the values of X, Y, M, and Z when the objective function expressed in equation (7) is minimized under the constraints of equations (8) to (12). The calculation unit 14 multiplies the found X by B to obtain a virtual partition size S. x This is calculated as the horizontal division size. The calculation unit 14 calculates the virtual division size S obtained by multiplying the calculated Y by B. y This is calculated as the vertical division size. Even when using this calculation method, it is possible to improve the accuracy of object detection.
[0065] In step S3, the centroid acquisition unit 12 calculated representative points using kernel density estimation. However, the centroid acquisition unit 12 may cluster the group of centroid coordinates and extract multiple representative points from the clusters. Alternatively, the centroid acquisition unit 12 may extract one average value from the group of centroid coordinates and use that average value as the representative point.
[0066] In step S2, the center of gravity acquisition unit 12 acquired the center of gravity of the object based on the bounding box, but the center of gravity acquisition unit 12 may acquire the center of gravity of the object using means other than the bounding box.
[0067] In step S3, the center of gravity acquisition unit 12 acquired a representative point of the center of gravity of each object class, but the center of gravity acquisition unit 12 does not need to acquire a representative point. In other words, step S3 may be omitted. In this case, in step S5, the calculation unit 14 may use the center of gravity of the object directly instead of a representative point to calculate the division size.
[0068] The calculation unit 14 may calculate the division size so that objects are not cut when dividing the image. For example, the calculation unit 14 may increase the value of β in the above formula (1) to Z c,i By increasing the weight of this factor, one can calculate the division size at which the object is not cut.
[0069] The method of calculating the division size using the above formulas (1) to (12) is merely an example, and other calculation methods may be used.
[0070] The detection unit 16 only needs to be able to perform object detection, and the form of the object detection model used is not limited.
[0071] The apparatus and method of this disclosure have the following configurations.
[0072] [1] An apparatus comprising: an image acquisition unit that acquires an image of an object; a size acquisition unit that acquires the object size of the object based on the image; a calculation unit that calculates a division size for dividing the image based on the object size; and a division unit that divides the image based on the division size.
[0073] [2] The apparatus according to [1], further comprising a center of gravity acquisition unit that acquires the center of gravity of the object based on the image, wherein the calculation unit calculates the division size based on the size of the object and the center of gravity.
[0074] [3] The apparatus according to [2], wherein the image acquisition unit acquires the image in which a rectangular bounding box surrounding the object is set, the size acquisition unit acquires the size of the object based on the bounding box, and the center of gravity acquisition unit acquires the center of gravity based on the bounding box.
[0075] [4] The apparatus according to [2] or [3], wherein the center of gravity acquisition unit acquires a representative point of the center of gravity of the object for each class of the object, and the calculation unit calculates the division size based on the size of the object and the representative point.
[0076] [5] The apparatus according to [4], wherein the center of gravity acquisition unit acquires the representative point based on the respective centers of gravity of a plurality of objects belonging to the same class.
[0077] [6] The apparatus according to any one of [1] to [5], wherein the calculation unit calculates the division size based on a virtual division size and a cutting size indicating the size by which the object protrudes from the divided images when the image is divided by the virtual division size.
[0078] [7] The apparatus according to [6], wherein the calculation unit calculates the virtual division size as the division size when the objective function relating the sum of the virtual division size and the cutting size is minimized.
[0079] [8] The apparatus according to [7], wherein the calculation unit calculates the virtual division size when the objective function is minimized using an optimization method.
[0080] [9] The apparatus according to any one of [2] to [8], wherein the calculation unit calculates the division size based on a virtual division size and a cutting size indicating the size by which the object extends beyond the divided image when the image is divided by the virtual division size, and the calculation unit calculates the cutting size based on the virtual division size, the size of the object, and the center of gravity.
[0081]
[10] A method of the apparatus comprising: acquiring an image in which an object is depicted; acquiring the object size of the object based on the image; calculating a division size for dividing the image based on the object size; and dividing the image based on the division size.
[0082] The block diagrams used in the description of the above embodiments show functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. The method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one device that is physically or logically coupled, or it may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wired or wireless connections). A functional block may be realized by combining the one or more devices with software.
[0083] Functions include, but are not limited to, judgment, decision, judgment, calculation, calculation, processing, derivation, investigation, exploration, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, assumption, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating (mapping), and assigning. For example, a functional block (configuration part) that enables transmission is called a transmitting unit or transmitter. In all cases, as mentioned above, the method of implementation is not particularly limited.
[0084] For example, the detection device 10 in one embodiment of the present disclosure may function as a computer that processes the method of the present disclosure. Figure 9 is a diagram showing an example of the hardware configuration of the device according to one embodiment of the present disclosure. The above-described detection device 10 may be physically configured as a computer device including a processor 1001, memory 1002, storage 1003, communication device 1004, input device 1005, output device 1006, and bus 1007.
[0085] In the following explanation, the term "device" can be replaced with "circuit," "device," or "unit." The hardware configuration of the detection device 10 may include one or more of the devices shown in the figure, or it may be configured to omit some of the devices.
[0086] Each function in the detection device 10 is realized by loading predetermined software (programs) onto hardware such as the processor 1001 or memory 1002, which causes the processor 1001 to perform calculations, control communication by the communication device 1004, and control at least one of data reading and writing in the memory 1002 and storage 1003.
[0087] The processor 1001 controls the entire computer, for example, by running an operating system. The processor 1001 may be composed of a central processing unit (CPU) that includes interfaces with peripheral devices, control devices, arithmetic units, and registers. For example, at least one of the functional parts of the detection device 10 described above may be implemented by the processor 1001.
[0088] The processor 1001 reads programs (program code), software modules, data, etc., from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes accordingly. The program used is one that causes the computer to execute at least a part of the operations described in the above embodiment. For example, at least one of the functional parts of the detection device 10 described above may be stored in the memory 1002 and implemented by a control program that runs on the processor 1001, and other functional blocks may be implemented similarly. The above processes have been described as being executed by one processor 1001, but they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may be transmitted from a network via a telecommunications line.
[0089] The memory 1002 is a computer-readable recording medium and may consist of at least one of the following: ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), and RAM (Random Access Memory). The memory 1002 may also be called a register, cache, or main memory. The memory 1002 can store executable programs (program code) and software modules for carrying out a detection method according to one embodiment of the present disclosure.
[0090] The storage 1003 is a computer-readable recording medium and may consist of at least one of the following: an optical disc such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital multipurpose disk, a Blu-ray® disc), a smart card, flash memory (e.g., a card, a stick, a key drive), a floppy® disk, and a magnetic strip. The storage 1003 may also be called an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, server, or other suitable medium including at least one of memory 1002 and storage 1003.
[0091] The communication device 1004 is hardware (transmitting / receiving device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as a network device, network controller, network card, or communication module, for example. The communication device 1004 may be configured to include, for example, a high-frequency switch, duplexer, filter, and frequency synthesizer in order to implement at least one of frequency division duplex (FDD) and time division duplex (TDD). For example, at least one of the functional parts of the detection device 10 described above may be implemented by the communication device 1004. The communication device 1004 may be implemented with a transmitting unit and a receiving unit that are physically or logically separated.
[0092] The input device 1005 is an input device that accepts input from an external source (e.g., a keyboard, mouse, microphone, switch, button, sensor, etc.). The output device 1006 is an output device that outputs to an external source (e.g., a display, speaker, LED lamp, etc.). The input device 1005 and the output device 1006 may be integrated (e.g., a touch panel).
[0093] The processor 1001 and the memory 1002, among other devices, are connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or different buses may be configured for each device.
[0094] The detection device 10 may be configured with hardware (processing circuits) such as a microprocessor, a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array), and some or all of each functional block may be realized by this hardware. For example, the processor 1001 may be implemented using at least one of these hardware components.
[0095] Information notification is not limited to the embodiments described herein and may be carried out by other means. For example, information notification may be carried out by physical layer signaling (e.g., DCI (Downlink Control Information), UCI (Uplink Control Information)), upper layer signaling (e.g., RRC (Radio Resource Control) signaling, MAC (Medium Access Control) signaling, broadcast information (MIB (Master Information Block), SIB (System Information Block))), other signals, or combinations thereof. RRC signaling may also be called RRC messages, and may be, for example, RRC Connection Setup messages, RRC Connection Reconfiguration messages, etc.
[0096] In the processing procedures, sequences, and flowcharts of each aspect / embodiment described herein, the order of processing may be changed as long as it is consistent with the original. For example, the methods described herein present the elements of various steps using an exemplary order and are not limited to the specific order presented.
[0097] Input and output information may be stored in a specific location (e.g., memory) or managed using a management table. Input and output information may be overwritten, updated, or appended to. Output information may be deleted. Input information may be transmitted to other devices.
[0098] The determination may be made by a value represented by one bit (0 or 1), by a boolean value (true or false), or by a numerical comparison (for example, a comparison with a predetermined value).
[0099] Each aspect / embodiment described herein may be used individually, in combination, or switched between as needed in practice. Notification of certain information (e.g., notification that "X is") is not limited to explicit notification, but may also be implicit (e.g., by not providing such notification).
[0100] Although the Disclosure has been described in detail above, it will be clear to those skilled in the art that the Disclosure is not limited to the embodiments described herein. The Disclosure may be implemented in modified and altered forms without departing from the intent and scope of the Disclosure as defined by the claims. Accordingly, the descriptions in the Disclosure are illustrative and not restrictive in any way.
[0101] Software should be interpreted broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, and functions, whether they are called software, firmware, middleware, microcode, or hardware description languages, or by any other name.
[0102] Software, instructions, and information may be transmitted or received via a transmission medium. For example, if software is transmitted from a website, server, or other remote source using at least one of wired technologies (such as coaxial cable, fiber optic cable, twisted pair, or digital subscriber line (DSL)) and wireless technologies (such as infrared or microwave), then at least one of these wired and wireless technologies is included in the definition of a transmission medium.
[0103] The information and signals described in this disclosure may be represented using any of the various different techniques. For example, the data, instructions, commands, information, signals, bits, symbols, and chips that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.
[0104] Terms used in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of the channel and symbol may be a signal (signaling). A signal may also be a message. A component carrier (CC) may also be called a carrier frequency, cell, or frequency carrier.
[0105] The information and parameters described in this disclosure may be expressed using absolute values, relative values from a given value, or other corresponding information. For example, wireless resources may be indicated by an index.
[0106] The names used for the parameters described above are not restrictive in any way. Furthermore, formulas and other mathematical expressions using these parameters may differ from those expressly disclosed in this disclosure. Various channels (e.g., PUCCH, PDCCH, etc.) and information elements can be identified by any suitable name, and therefore, the various names assigned to these various channels and information elements are not restrictive in any way.
[0107] In this disclosure, terms such as "Mobile Station (MS)," "user terminal," "User Equipment (UE)," and "terminal" may be used interchangeably.
[0108] A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or several other appropriate terms.
[0109] As used in this disclosure, the terms “determining” and “determining” may encompass a wide variety of actions. “Determining” may be considered, for example, judging, calculating, computing, processing, deriving, investigating, looking up, search, inquiry (e.g., searching in tables, databases, or other data structures), ascertaining, etc. “Determining” may be considered receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in memory), etc. “Determining” may be considered resolving, selecting, choosing, establishing, comparing, etc. In other words, “determining” and “determining” may include being considered any action related to “determining” and “determining.” "Judgment (decision)" can be replaced with "assuming," "expecting," or "considering."
[0110] The terms “connected,” “coupled,” or any variation thereof, mean any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are “connected” or “coupled” with each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, “connection” may be reinterpreted as “access.” Where “connected” or “coupled” is used in this disclosure, two elements may be considered to be “connected” or “coupled” with each other using at least one of one or more wires, cables, and printed electrical connections, and may also be considered to be “connected” or “coupled” with each other using, in some non-limiting and non-exclusive examples, electromagnetic energy having wavelengths in the radio frequency domain, microwave domain, and optical (both visible and invisible) domain.
[0111] As used in this disclosure, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based on at least."
[0112] Any reference to elements using designations such as “first” and “second” as used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient way to distinguish between two or more elements. Accordingly, references to the first and second elements do not imply that only two elements may be employed, nor that the first element must precede the second element in any way.
[0113] Where the terms “include,” “including,” and variations thereof are used in this disclosure, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to mean exclusive OR.
[0114] In this disclosure, if articles are added through translation, such as "a," "an," and "the" in English, this disclosure may include the fact that the noun following these articles is plural.
[0115] In this disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "combine" may also be interpreted similarly to "different."
[0116] 10...Detection device, 11...Image acquisition unit, 12...Center of gravity acquisition unit, 13...Size acquisition unit, 14...Calculation unit, 15...Division unit, 16...Detection unit, BB...Bounding box, CL...Class, CP...Cutting portion, OB...Object, P...Center of gravity, RP...Representative point.
Claims
1. An apparatus comprising: an image acquisition unit that acquires an image of an object; a size acquisition unit that acquires the object size of the object based on the image; a calculation unit that calculates a division size for dividing the image based on the object size; and a division unit that divides the image based on the division size.
2. The apparatus according to claim 1, further comprising a center of gravity acquisition unit that acquires the center of gravity of the object based on the image, wherein the calculation unit calculates the division size based on the size of the object and the center of gravity.
3. The apparatus according to claim 2, wherein the image acquisition unit acquires the image in which a rectangular bounding box surrounding the object is set, the size acquisition unit acquires the size of the object based on the bounding box, and the center of gravity acquisition unit acquires the center of gravity based on the bounding box.
4. The apparatus according to claim 2, wherein the center of gravity acquisition unit acquires a representative point of the center of gravity of the object for each class of the object, and the calculation unit calculates the division size based on the size of the object and the representative point.
5. The apparatus according to claim 4, wherein the center of gravity acquisition unit acquires the representative point based on the respective centers of gravity of a plurality of objects belonging to the same class.
6. The apparatus according to claim 1, wherein the calculation unit calculates the division size based on a virtual division size and a cutting size indicating the size by which the object protrudes from the divided image when the image is divided by the virtual division size.
7. The apparatus according to claim 6, wherein the calculation unit calculates the virtual division size as the division size when the objective function relating the sum of the virtual division size and the cutting size is minimized.
8. The apparatus according to claim 7, wherein the calculation unit calculates the virtual division size when the objective function is minimized using an optimization method.
9. The apparatus according to claim 2, wherein the calculation unit calculates the division size based on a virtual division size and a cutting size indicating the size by which the object extends beyond the divided image when the image is divided by the virtual division size, and the calculation unit calculates the cutting size based on the virtual division size, the size of the object, and the center of gravity.
10. A method of an apparatus comprising: acquiring an image in which an object is depicted; acquiring the object size of the object based on the image; calculating division sizes for dividing the image based on the object size; and dividing the image based on the division sizes.
