Material recognition method, material sorting method, material recognition device, and apparatus

Abstract

The present disclosure relates to the technical field of raw coal sorting, and particularly relates to a material identification method, a material sorting method, a material identification device and equipment. The material identification method comprises the following steps: obtaining a target image of a material to be identified; determining the contour of the material to be identified based on the target image; inputting the target image into a target identification model to determine a plurality of identification boxes included in the target image and corresponding material types; and determining the type of the material to be identified based on the contour, the plurality of identification boxes and the corresponding material types. The problem of spatial misplacement can be avoided, the limitation of a single feature can be avoided, and impurity interference can be excluded, thereby helping to improve the accuracy and reliability of overall material identification, and enhancing the environmental interference resistance of material identification.

Description

Technical Field

[0001] This disclosure relates to the field of raw coal sorting technology, specifically to a material identification method, a material sorting method, a material identification device, a material sorting equipment, and a computer-readable storage medium. Background Technology

[0002] In the field of raw coal sorting, accurate identification and separation of coal and gangue are core steps in improving coal quality, increasing sorting efficiency, and reducing resource waste. Related technologies often employ a combination of X-ray detection and industrial camera image analysis to identify materials.

[0003] However, in actual processing, X-rays are not sensitive to the differences between high-density coal and low-density gangue (such as white sandstone), which can easily lead to misjudgments. Furthermore, industrial camera images are easily affected by external environmental factors such as coal slime coverage and changes in lighting, making feature extraction difficult. In addition, differences in the installation of the X-ray source and the camera can cause spatial misalignment, making it easy for interfering targets to be mixed into the intercepted area. Coupled with the fact that identification is based on a single surface feature, the final result is low identification accuracy, affecting the sorting precision. Summary of the Invention

[0004] To overcome the problems existing in the related technologies, an exemplary embodiment of this disclosure provides a material identification method, including: acquiring a target image of a material to be identified; determining the outline of the material to be identified based on the target image; inputting the target image into a target identification model to determine multiple identification boxes included in the target image and the corresponding material types; and determining the type of the material to be identified based on the outline, the multiple identification boxes, and the corresponding material types.

[0005] In some embodiments, the type of the material to be identified is determined based on the contour, multiple recognition boxes, and the corresponding material type, including: if a first recognition box exists among the multiple recognition boxes, the type of the material to be identified is determined to be a first type, wherein the first recognition box is a recognition box that includes the entire area of ​​the contour and the corresponding material type is the first type; if a first recognition box does not exist among the multiple recognition boxes, the type of the material to be identified is determined based on the area of ​​a second recognition box, wherein the second recognition box is a recognition box that includes a partial area of ​​the contour and the corresponding material type is the first type.

[0006] In some embodiments, determining the type of material to be identified based on the area of ​​the second identification box includes: determining the intersection area between the second identification box and the contour; if the ratio of the intersection area to the area of ​​the contour is greater than a target threshold, then the type of the material to be identified is determined to be a first type.

[0007] In some embodiments, determining the type of material to be identified based on the area of ​​the second identification box further includes: if the ratio of the intersection area to the area of ​​the contour is less than or equal to the target threshold, then the type of the material to be identified is determined to be the second type.

[0008] In some embodiments, determining the type of the material to be identified based on the contour, multiple recognition boxes, and the corresponding material types further includes: if the material types corresponding to the multiple recognition boxes are all of the second type, then determining the type of the material to be identified as the second type.

[0009] In some embodiments, the training process of the target recognition model includes: acquiring multiple image samples and multiple recognition box information corresponding to each image sample, wherein the material type corresponding to the multiple recognition box information includes a first type and / or a second type; and training an initial recognition model based on the multiple image samples and the multiple recognition box information corresponding to each image sample to obtain the target recognition model.

[0010] Secondly, this disclosure also provides a material sorting method, comprising: determining the type of material by means of the material identification method provided in any of the above aspects; and sorting the material according to the type.

[0011] Thirdly, this disclosure also provides a material identification device, comprising: an acquisition module for acquiring a target image of a material to be identified; a processing module for determining the outline of the material to be identified based on the target image; and an identification module for inputting the target image into a target identification model, determining multiple identification boxes included in the target image and the corresponding material types, and determining the type of the material to be identified based on the outline, the multiple identification boxes, and the corresponding material types.

[0012] Fourthly, this disclosure also provides a material sorting device, comprising: a material identification device provided in any of the above aspects, for determining the type of material; and a sorting mechanism for sorting materials according to their type.

[0013] Fifthly, this disclosure also provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to perform the material identification method or the material sorting method provided in any of the above aspects.

[0014] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure.

[0015] The technical solutions provided by the embodiments of this disclosure can include the following beneficial effects: According to the material identification method provided by this disclosure, by acquiring the target image of the material to be identified, the integrity of the information of the material to be identified can be guaranteed, laying a reliable foundation for subsequent material identification through the target identification model, and avoiding the occurrence of missed identification due to missing features. By determining the contour of the material to be identified in the target image, the spatial position of the material can be clearly defined, avoiding the problem of spatial misalignment. By identifying multiple bounding boxes and corresponding material types contained in the target image through the target identification model, and then combining the contour position to jointly determine the type of material to be identified, the limitations of a single feature can be effectively avoided, and interference from impurities can be eliminated, thereby helping to improve the accuracy and reliability of the overall material identification and enhancing the material identification's resistance to environmental interference. Attached Figure Description

[0016] This disclosure can be better understood by describing exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, in which: Figure 1 This is a schematic diagram illustrating the device layout according to an exemplary embodiment disclosed in a publication; Figure 2 This is a schematic diagram illustrating an image of a mineral according to an exemplary embodiment disclosed in a publication; Figure 3 This is a schematic diagram illustrating another mineral image according to an exemplary embodiment of a disclosed document; Figure 4 This is a flowchart illustrating a material identification method according to an exemplary embodiment of a published document; Figure 5 This is a schematic diagram illustrating a standard embodiment of a published model. Figure 6 This is a flowchart illustrating another material identification method according to an exemplary embodiment of a published document; Figure 7 This is a schematic diagram illustrating yet another mineral image according to an exemplary embodiment of a published document; Figure 8 This is a schematic diagram illustrating another mineral image according to an exemplary embodiment of a disclosed publication; Figure 9 This is a schematic flowchart illustrating a material sorting method according to an exemplary embodiment of a published document; Figure 10 This is a schematic diagram of the structure of a material identification device according to an exemplary embodiment disclosed in a publication; Figure 11 This is a schematic diagram of the structure of a material sorting device according to an exemplary embodiment disclosed in a book. Detailed Implementation

[0017] The following describes specific embodiments of this disclosure. It should be noted that, in order to provide a concise description, this specification cannot exhaustively describe all features of the actual embodiments. It should be understood that, in the actual implementation of any embodiment, just as in any engineering or design project, various specific decisions are often made to achieve the developer's specific goals and to meet system-related or business-related constraints, and this can change from one embodiment to another. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this disclosure, changes in design, manufacturing, or production based on the technical content disclosed in this disclosure are merely conventional technical means and should not be construed as insufficient content of this disclosure.

[0018] Unless otherwise defined, the technical or scientific terms used in this disclosure shall have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms “first,” “second,” and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. The terms “a” or “one,” etc., do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms “comprising,” “including,” etc., mean that the element or object preceding “comprising” or “including” encompasses the elements or objects listed following “comprising” or “including” and their equivalents, and do not exclude other elements or objects. The terms “connected,” “linked,” etc., are not limited to physical or mechanical connections, nor are they limited to direct or indirect connections.

[0019] In related technologies, a combination of X-ray detection and industrial camera image analysis is often used to identify materials. For example, firstly, X-rays are used to penetrate the ore, and a grayscale image is generated based on its density differences. Then, the image is segmented to obtain the contour information of the ore and its bounding rectangle is calculated. Next, a line-scan industrial camera installed at a certain distance from the X-ray source is used to acquire an image of the ore's surface. Finally, based on the bounding rectangle obtained from the aforementioned X-ray image, the surface image acquired by the industrial camera is cropped (usually requiring size scaling for matching), and the cropped image is used for target recognition to extract the ore's surface features and thus determine its properties.

[0020] However, in actual processing, the above-mentioned scheme faces many challenges that affect recognition accuracy, mainly in the following aspects: 1. Core Identification Challenges: X-ray inspection is insensitive to the density / atomic number differences between high-density coal and certain low-density gangue (such as white sandstone), easily leading to misjudgments. Simultaneously, images captured by industrial cameras are easily affected by factors such as coal slime coverage and changes in lighting, resulting in blurred key features such as ore surface texture and color, making effective extraction difficult and further exacerbating the identification challenges.

[0021] 2. Spatial Misalignment and Interference Introduction: Due to the different installation heights and angles of the X-ray source and the industrial camera, the ore projection outline obtained in the X-ray image does not spatially match the surface image captured by the industrial camera. Therefore, the industrial camera image area (i.e., the region of interest) extracted based on the X-ray outline often contains not only the target ore but may also include parts or whole pieces of other ores, small particles of gangue, or coal, etc., which are interfering targets.

[0022] 3. Impact of interfering targets: When an image is captured containing interfering targets, irrelevant feature information will inevitably be introduced during subsequent surface feature extraction and target recognition processes. This will cause confusion in the target recognition algorithm, making it impossible to accurately determine the true properties of the main target ore in the captured area. As a result, sorting errors will occur, leading to resource waste or substandard coal product quality.

[0023] 4. Limited Feature Recognition: Existing technologies often rely on a single or a few surface features extracted from the cropped image when making the final judgment, lacking a comprehensive consideration of the multi-dimensional information of the ore, which limits the robustness and accuracy of the judgment model.

[0024] For example, such as Figure 1 As shown, when the ore (circle in the figure) is positioned near the center of the material conveying mechanism, the ore position BD acquired by X-rays does not coincide with the position AC acquired by the industrial camera. This results in the inability to accurately extract the camera contour of the ore based on the X-ray profile. Figure 2 Taking an industrial image of ore as an example, the gray solid line represents the detected ray contour. This image shows that the ray contour cannot protect the entire ore area in the image, and the missing portion is crucial for identifying the ore type. If only the area within the ray contour is examined, the ore might be identified as gangue. However, the silver reflective area on the left indicates coal, because a collision occurred there. When there are no obvious distinguishing features on the surface, the collision area often reveals the accurate characteristics of the ore. Therefore, directly using ray contours to extract ore targets from camera images easily leads to the loss of feature areas.

[0025] However, if the ray outline is expanded to extract the ore region, although the resulting image can contain all the information about the ore, it will also include small ore particles adjacent to the ore, such as... Figure 3As shown. Therefore, recognition through the augmented image inevitably introduces irrelevant feature information, causing confusion in the target recognition algorithm. This makes it impossible to accurately determine the true properties of the main target ore within the intercepted area, thereby increasing the algorithm's time consumption and affecting its use in the field of high-speed ore sorting.

[0026] To address the aforementioned problems, this disclosure provides a material identification method. For example... Figure 4 As shown, the material identification method may include the following steps: Step S110: Obtain the target image of the material to be identified.

[0027] The target image is an image captured by an industrial camera that contains complete information about the material to be identified. To ensure the accuracy of material identification, the target image of the material to be identified is acquired so that missed identifications can be avoided during subsequent identification, laying the foundation for improving identification accuracy in the future.

[0028] Step S120: Determine the outline of the material to be identified based on the target image.

[0029] Based on the mapping relationship between the target image and the ray grayscale image, the outline of the material to be sorted in the target image can be effectively identified, thereby determining the position of the material to be identified in the target image, which helps to provide a reliable positioning basis for subsequent feature recognition.

[0030] Step S130: Input the target image into the target recognition model to determine the multiple recognition boxes included in the target image and the corresponding material types.

[0031] Because the target image may contain adjacent impurities in addition to the material to be identified, it can easily interfere with the identification of the target material type. Therefore, to improve the accuracy of material identification, the target image is input into a target recognition model. The model identifies the material types and corresponding locations that may be included in the target image and marks them with corresponding bounding boxes. For example, a correspondence between the color of the bounding box and the material type is established in advance, and then a bounding box of the corresponding color is used to mark the material type based on the identified material type. The target recognition model is a pre-trained recognition model that can identify not only the material type corresponding to global image features but also the material type corresponding to local image features. The output recognition result includes multiple bounding boxes and their corresponding material types. The material types corresponding to multiple bounding boxes can be the same or multiple. The specific material type depends on the actual image features of the target image and the recognition capability of the target recognition model.

[0032] In some application scenarios, such as Figure 5As shown, the target image is detected by the target recognition model, and four areas containing materials are identified. Then, according to the type of material identified, the corresponding recognition boxes are marked to obtain four recognition boxes in the figure (a large rectangle containing the entire image and three small rectangles in the right area of ​​the figure).

[0033] Step S140: Determine the type of material to be identified based on the outline, multiple recognition boxes, and the corresponding material type.

[0034] By analyzing the contours, the location of the material to be identified within the target image can be determined. By identifying the bounding boxes and their corresponding material types, the material type corresponding to the image regions marked by each bounding box in the target image can be determined. By matching the image regions marked by each bounding box with the regions containing the contours, the bounding boxes that best fit the contour can be selected, and the corresponding material types can be used as the type of material to be identified. This effectively reduces interference from impurity areas and ensures the reliability and accuracy of material type determination.

[0035] According to the material identification method provided in this disclosure, by acquiring the target image of the material to be identified, the integrity of the material's information can be ensured, laying a reliable foundation for subsequent material identification through a target recognition model and avoiding missed identification due to missing features. By determining the contour of the material to be identified in the target image, the spatial position of the material can be clearly defined, avoiding spatial misalignment. By identifying multiple bounding boxes and their corresponding material types contained in the target image through the target recognition model, and then combining this with the contour position to jointly determine the material type to be identified, the limitations of single features can be effectively avoided, and interference from impurities can be eliminated. This helps to improve the overall accuracy and reliability of material identification and enhances the material identification's resistance to environmental interference.

[0036] In some embodiments, such as Figure 6 As shown, step S140 above may include the following steps: Step S141: If a first identification box exists in multiple identification boxes, then the type of the material to be identified is determined to be the first type.

[0037] The first identification box is a box that includes the entire outline area and corresponds to the first type of material. Since the first identification box covers the entire area where the material to be sorted is located, the material type corresponding to the first identification box can be directly used as the type of the material to be identified, thereby helping to improve the material identification efficiency.

[0038] Step S142: If the first identification box is not present in any of the multiple identification boxes, the type of material to be identified is determined based on the area of ​​the second identification box.

[0039] The second recognition box is a box that includes a local contour region and corresponds to the first type of material. The first type is the desired material type. If the first recognition box is absent among multiple recognition boxes, but the second recognition box is present, it indicates that the target image contains material image features corresponding to the first type. Therefore, to avoid misidentification, the overlap between the location of the second recognition box and the location of the contour region is determined based on the recognition area of ​​the second recognition box. This helps determine whether the presence of the second recognition box is due to environmental interference or the structure of the material itself, thereby determining the type of material to be identified.

[0040] In some examples, step a2 above may include the following steps: Step a21: Determine the intersection area between the second recognition box and the contour; Step a22: If the ratio of the intersection area to the area of ​​the contour is greater than the target threshold, then the type of the material to be identified is determined to be the first type.

[0041] Specifically, the larger the intersection area between the second identification box and the contour, the higher the reliability of its representation of the material type corresponding to the contour; conversely, the smaller the intersection area, the lower the reliability of its representation of the material type corresponding to the contour. Therefore, to determine whether the material type corresponding to the second identification box can represent the material type of the material to be identified, a target threshold is predetermined. The target threshold is the minimum critical ratio used to measure whether the area proportion is effective. For example, the target threshold could be 1 / 3.

[0042] If the ratio of the intersection area to the contour area is greater than the target threshold, it indicates that the overlap between the second recognition box and the contour is high. The existence of the second recognition box is based on the surface image features of the material to be identified, and is not caused by environmental interference factors. Therefore, the second recognition box can identify the contour, and the first type corresponding to the second recognition box can be used as the type of the material to be identified, thereby effectively avoiding the occurrence of missed identification.

[0043] In other examples, step a2 above may also include the following steps: Step a23: If the ratio of the intersection area to the area of ​​the contour is less than or equal to the target threshold, then the type of the material to be identified is determined to be the second type.

[0044] If the ratio of the intersection area to the contour area is less than or equal to the target threshold, it indicates that the overlap between the second identification box and the contour is relatively low. The existence of the second identification box may be due to environmental interference factors. Therefore, the type of the material to be identified is determined to be the second type to avoid misidentification.

[0045] In other embodiments, step S140 may further include the following steps: Step S143: If the material type corresponding to multiple recognition boxes is the second type, then the type of the material to be recognized is determined to be the second type.

[0046] If multiple recognition boxes correspond to the same material type (Type II), it indicates that there is no ambiguity in determining the type of the material to be identified. Therefore, Type II can be directly used as the type of the material to be identified, which helps to improve the efficiency of determination.

[0047] In some other embodiments, if the target image is identified by the target recognition model but no recognition box is obtained, the type of the material to be identified is classified as the first type to prevent waste.

[0048] The training process of the target recognition model will be explained in detail below: Step b1: Obtain multiple image samples and multiple recognition bounding boxes corresponding to each image sample; Step b2: Based on multiple image samples and the information of multiple recognition boxes corresponding to each image sample, train the initial recognition model to obtain the target recognition model.

[0049] To enable the trained target recognition model to simultaneously output multiple bounding boxes and their corresponding material types, the required image samples and corresponding bounding box information are pre-constructed. Specifically, the annotations for each image sample must cover all possible material regions in the image, and the material types corresponding to the multiple bounding boxes must fully encompass the first type (target sorting material, such as coal) and / or the second type (non-target material, such as gangue), ensuring that the initial recognition model can learn the feature differences between the two material types.

[0050] When there are no obvious distinguishing features on the surface of the ore, the exposed areas after impact often reveal the true properties of the ore. For example, in some application scenarios, such as... Figure 7 The ore image shown, identified solely based on the area within the ray outline, is easily misclassified as coal; however, by examining the exposed areas at the points of impact, it can be accurately determined to be gangue. For example... Figure 8 The ore image shown can easily be misidentified as gangue if only the area inside the ray outline is observed, but it can be identified as coal by using the high brightness feature of the exposed area caused by impact.

[0051] Therefore, to ensure the reliability of the target recognition model, a "global + local" dual-dimensional annotation strategy is adopted for the same image sample: on the one hand, a global recognition box corresponding to the material type is annotated for the complete material area (such as annotating the "coal" type recognition box for a complete coal block); on the other hand, a local recognition box of the same material type is annotated for local areas with significant material surface features (such as the glossy area of ​​coal, the texture area of ​​gangue, the exposed area of ​​ore, etc.).

[0052] This annotation strategy ensures that the initial recognition model learns both the complete global surface features of the material and discriminative local surface features during training, making the learned feature information more comprehensive, detailed, and reliable. The resulting target recognition model can simultaneously and accurately identify both global material regions and local feature regions in an image, outputting multi-dimensional bounding boxes and their corresponding types. This provides sufficient and reliable recognition data for subsequent accurate matching and determination based on the material contour.

[0053] In other embodiments, the material identification method can be obtained through a large model identification process. That is, the large model includes a target identification model and a type determination module. Through this large model, the type of the material to be identified can be directly determined based on the acquired target image using the material identification method provided in this disclosure, thereby greatly shortening the identification time and improving the identification efficiency. This helps to promote the overall process of material identification, facilitates the improvement of material sorting efficiency, and enhances material sorting performance.

[0054] In some optional application scenarios, taking coal as the first type and gangue as the second type as an example, the process of identifying materials can be as follows: Acquire a target image of the material to be identified, and determine the outline of the material to be identified based on the target image.

[0055] The target image is input into the target recognition model to determine the multiple recognition boxes included in the target image and the corresponding material types.

[0056] If a first identification box exists among multiple identification boxes, then the type of material to be identified is determined to be coal. The first identification box is an identification box that encompasses the entire outline area and corresponds to the material type of coal.

[0057] If multiple identification boxes contain the entire area of ​​the outline, and the corresponding type is gangue, then the area of ​​the second identification box is determined. The second identification box is one that contains a partial area of ​​the outline and corresponds to the material type as coal. The intersection area between the second identification box and the outline is determined. If the ratio of the intersection area to the outline area is greater than a target threshold (1 / 3), then the type of material to be identified is determined to be coal. If the ratio of the intersection area to the outline area is less than or equal to the target threshold (1 / 3), then the type of material to be identified is determined to be gangue.

[0058] Alternatively, if none of the multiple identification boxes contain the entire area of ​​the contour, but a second identification box exists, the intersection area between the second identification box and the contour is determined. If the ratio of the intersection area to the area of ​​the contour is greater than a target threshold (1 / 3), the type of material to be identified is determined to be coal. If the ratio of the intersection area to the area of ​​the contour is less than or equal to the target threshold (1 / 3), the type of material to be identified is determined to be gangue.

[0059] If the material type corresponding to multiple identification boxes is gangue, then the material to be identified is determined to be gangue.

[0060] If no material type is identified, the material to be identified is determined to be coal in order to avoid resource waste.

[0061] Based on the same inventive concept, this disclosure also provides a material sorting method. For example... Figure 9 As shown, the material sorting method may include the following steps: Step S210: Determine the type of material using a material identification method. The material identification method can be any of the material identification methods provided in this disclosure.

[0062] Step S220: Sort materials according to type.

[0063] According to the material sorting method provided in this disclosure, the material type can be determined by the material identification method provided in this disclosure, which can ensure the accuracy and efficiency of material identification. Thus, the materials can be sorted according to the identified type, which can effectively improve the material sorting quality, shorten the sorting process, and ensure the material sorting performance.

[0064] Based on the same inventive concept, this disclosure also provides a material identification device. For example... Figure 10 As shown, the material identification device 300 may include: The acquisition module 310 is used to acquire a target image of the material to be identified; Processing module 320 is used to determine the outline of the material to be identified based on the target image; The recognition module 330 is used to input the target image into the target recognition model, determine the multiple recognition boxes included in the target image and the corresponding material types, and determine the type of the material to be recognized based on the contour, the multiple recognition boxes and the corresponding material types.

[0065] In some embodiments, the identification module 330 may include: a first identification unit, configured to determine the type of the material to be identified as a first type if a first identification box exists among multiple identification boxes, wherein the first identification box is an identification box that includes the entire outline area and the corresponding material type is the first type; and a second identification unit, configured to determine the type of the material to be identified based on the area of ​​a second identification box if a first identification box does not exist among multiple identification boxes, wherein the second identification box is an identification box that includes a partial outline area and the corresponding material type is the first type.

[0066] In some embodiments, the second identification unit may include: a first determining unit, configured to determine the intersection area between the second identification box and the contour; and a second determining unit, configured to determine the type of the material to be identified as a first type if the ratio of the intersection area to the area of ​​the contour is greater than a target threshold.

[0067] In some embodiments, the second identification unit may further include a third determination unit, configured to determine the type of the material to be identified as the second type if the ratio of the intersection area to the area of ​​the contour is less than or equal to the target threshold.

[0068] In some embodiments, the identification module 330 may further include: a third identification unit, configured to determine that the type of the material to be identified is the second type if the material types corresponding to the multiple identification boxes are all of the second type.

[0069] In some embodiments, the training apparatus for the target recognition model may include: a sample acquisition module, configured to acquire multiple image samples and multiple recognition box information corresponding to each image sample, wherein the material type corresponding to the multiple recognition box information includes a first type and / or a second type; and a training module, configured to train an initial recognition model based on the multiple image samples and the multiple recognition box information corresponding to each image sample to obtain the target recognition model.

[0070] Regarding the material identification device in the above embodiments, the specific way in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated here.

[0071] Based on the same inventive concept, this disclosure also provides a material sorting device. For example... Figure 11 As shown, the material sorting equipment may include: a material identification device 300 provided in this disclosure, used to determine the type of material; and a sorting mechanism 400, used to sort materials according to their type.

[0072] Based on the same inventive concept, this disclosure also provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the material identification method or the material sorting method of any of the foregoing embodiments.

[0073] This disclosure uses specific terms to describe embodiments of the present disclosure. Terms such as "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of the present disclosure. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Furthermore, certain features, structures, or characteristics in one or more embodiments of the present disclosure can be appropriately combined.

[0074] In the context of this disclosure, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0075] Similarly, it should be noted that, in order to simplify the description of this disclosure and thus aid in the understanding of one or more embodiments, the foregoing description of embodiments of this disclosure may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this disclosure method does not imply that the subject matter of this disclosure requires more features than the features claimed. In fact, the embodiments contain fewer features than all the features of the single embodiments disclosed above.

[0076] The basic concepts have been described above. It is obvious that the above disclosure is merely illustrative and does not constitute a limitation of this disclosure. Although not explicitly stated herein, various modifications, improvements, and corrections may be made to this disclosure by those skilled in the art. Such modifications, improvements, and corrections are suggested in this disclosure and therefore remain within the spirit and scope of the embodiments of this disclosure.

Claims

1. A material identification method, comprising: Acquire a target image of the material to be identified; Based on the target image, the outline of the material to be identified is determined; The target image is input into the target recognition model to determine the multiple recognition boxes included in the target image and the corresponding material types; Based on the contour, multiple recognition boxes, and corresponding material types, the type of the material to be identified is determined, including: if a first recognition box exists among the multiple recognition boxes, the type of the material to be identified is determined to be a first type, wherein the first recognition box is a recognition box that includes the entire area of ​​the contour and the corresponding material type is the first type; if the first recognition box does not exist among the multiple recognition boxes, the type of the material to be identified is determined based on the area of ​​a second recognition box, wherein the second recognition box is a recognition box that includes a partial area of ​​the contour and the corresponding material type is the first type; The step of determining the type of the material to be identified based on the area of ​​the second identification box includes: determining the intersection area between the second identification box and the contour; if the ratio of the intersection area to the area of ​​the contour is greater than a target threshold, then the type of the material to be identified is determined to be the first type; if the ratio of the intersection area to the area of ​​the contour is less than or equal to the target threshold, then the type of the material to be identified is determined to be the second type.

2. The material identification method according to claim 1, wherein, The step of determining the type of the material to be identified based on the contour, multiple recognition boxes, and corresponding material types further includes: If the material type corresponding to multiple recognition boxes is the second type, then the type of the material to be recognized is determined to be the second type.

3. The material identification method according to claim 1, wherein, The training process of the target recognition model includes: Acquire multiple image samples and multiple recognition box information corresponding to each image sample, wherein the material types corresponding to the multiple recognition box information include a first type and / or a second type; Based on the multiple image samples and the multiple recognition box information corresponding to each image sample, an initial recognition model is trained to obtain the target recognition model.

4. A material sorting method, comprising: The material identification method according to any one of claims 1-3 is used to determine the type of material; The materials are sorted according to their type.

5. A material identification device for determining the type of a material to be identified using the material identification method according to any one of claims 1-3, comprising: The acquisition module is used to acquire the target image of the material to be identified; The processing module is used to determine the outline of the material to be identified based on the target image; The recognition module is used to input the target image into the target recognition model, determine multiple recognition boxes and corresponding material types included in the target image, and determine the type of the material to be recognized based on the contour, the multiple recognition boxes and the corresponding material types.

6. A material sorting device, comprising: The material identification device according to claim 5 is used to determine the type of material; A sorting mechanism for sorting the materials according to the type.

7. A computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a computer to perform the material identification method of any one of claims 1 to 3 or the material sorting method of claim 4.

Images