Defect detection method, device and equipment of battery pole piece, and storage medium

Abstract

The application discloses a battery pole piece defect detection method, device, apparatus and storage medium, wherein the method comprises: acquiring a first image and a second image obtained by shooting a battery pole piece to be detected, wherein a shooting parameter of the first image is a preset parameter suitable for shooting a pole plate region of the battery pole piece; and a shooting parameter of the second image is a preset parameter suitable for shooting a tab region of the battery pole piece; performing image detection on the first image and the second image to obtain the pole plate region and the tab region; and performing defect detection on the pole plate region and the tab region to obtain a defect detection result of the battery pole piece.

Description

Technical Field

[0001] This application relates to the field of image detection technology, and to, but is not limited to, a method, apparatus, device, and storage medium for detecting defects in battery electrodes. Background Technology

[0002] Electrodes are crucial components in the assembly of batteries, such as lithium-ion batteries. During the production of lithium-ion battery electrodes, defects such as exposed foil, dark spots, bright spots, and material loss can occur due to the coating machine and rolling mill. All electrodes must be inspected before battery lamination and assembly to identify and separately store and dispose of defective, substandard, and good electrodes. Currently, defect detection of lithium-ion battery electrodes is primarily done through manual visual inspection. The reliability, stability, and efficiency of manual inspection cannot be effectively controlled, and the high cost and labor-intensive nature of existing labor directly restrict the low-cost production of lithium-ion batteries. Summary of the Invention

[0003] In view of this, embodiments of this application provide a method, apparatus, device, and storage medium for detecting defects in battery electrodes.

[0004] In a first aspect, embodiments of this application provide a method for defect detection of a battery electrode, the electrode comprising an electrode plate and a tab, the method comprising: acquiring a first image and a second image obtained by photographing the battery electrode to be detected, wherein the photographing parameters of the first image are preset parameters suitable for photographing the electrode plate region of the battery electrode; the photographing parameters of the second image are preset parameters suitable for photographing the tab region of the battery electrode; performing image detection on the first image and the second image to obtain an electrode plate region and a tab region; and performing defect detection on the electrode plate region and the tab region to obtain a defect detection result of the battery electrode.

[0005] In some embodiments, performing image detection on the first image and the second image to obtain the electrode plate region and the electrode tab region includes: using a target detection algorithm to perform image detection on the first image and the second image to obtain the electrode plate region and the electrode tab region.

[0006] In this way, by using object detection algorithms to perform image detection on the first and second images, it can be better applied to the production line process and improve production efficiency.

[0007] In some embodiments, the step of performing image detection on the first image and the second image to obtain the electrode plate region and the electrode tab region includes: performing image detection on the first image to obtain an electrode plate detection frame; performing image detection on the second image to obtain an electrode tab detection frame; and cropping the electrode plate detection frame and the electrode tab detection frame respectively to obtain the electrode plate region and the electrode tab region.

[0008] In this way, by performing image detection on the first image, a plate detection box is obtained; by performing image detection on the second image, a tab detection box is obtained; then the plate detection box and the tab detection box are cropped out for subsequent defect detection, which can reduce the interference of the background in the first and second images on the defect detection results.

[0009] In some embodiments, the step of using a target detection algorithm to perform image detection on the first image and the second image to obtain the electrode region and the tab region includes: stitching the first image and the second image to obtain a battery electrode image; using the target detection algorithm to simultaneously perform electrode detection and tab detection on the first image and the second image in the battery electrode image to obtain at least one electrode detection box and at least one tab detection box; and determining the electrode region and the tab region in the at least one electrode detection box and the at least one tab detection box based on the relative positional relationship between the first image and the second image in the battery electrode image.

[0010] In this way, by stitching the first image and the second image together, a battery electrode image is obtained. Then, image detection is performed on the battery electrode image to obtain the plate area and the tab area. Finally, defect detection is performed on the plate area and the tab area, thereby realizing end-to-end detection of battery electrode defects.

[0011] In some embodiments, the step of separately cutting the electrode plate detection frame and the electrode tab detection frame to obtain the electrode plate region and the electrode tab region includes: expanding the electrode plate detection frame and the electrode tab detection frame outward to obtain the electrode plate area to be processed and the electrode tab area to be processed; and separately cutting the electrode plate area to be processed and the electrode tab area to be processed to obtain the electrode plate region and the electrode tab region.

[0012] In this way, by expanding the electrode plate detection frame and the electrode tab detection frame to obtain the electrode plate to be processed area and the electrode tab to be processed area, the electrode plate to be processed area and the electrode tab to be processed area can include defects at the edges of the electrode plate and the electrode tab, thereby improving the accuracy of defect detection.

[0013] In some embodiments, the defect detection of the electrode plate region and the tab region to obtain the defect detection result of the battery electrode includes: classifying the electrode plate region and the tab region respectively using a classification algorithm to obtain the defect detection result of the battery electrode.

[0014] In this way, by using classification algorithms to classify the electrode plate area and the electrode tab area, the defect detection results can be obtained. This can improve the accuracy of defect detection and be better applied to the fast-paced production process of the production line, thereby improving production efficiency.

[0015] In some embodiments, the electrode plate is classified into multiple types including: damaged, exposed white, wrinkled, missing material, foreign matter, color difference, crack, and normal; the tab is classified into multiple types including: damaged, misaligned, and normal.

[0016] The step of classifying the plate region and the tab region using classification algorithms to obtain the defect detection result of the battery electrode includes: classifying the plate region and the tab region using classification algorithms to obtain the confidence level of each classification type of the plate and the confidence level of each classification type of the tab; and determining the defect detection result of the battery electrode based on the confidence level of each classification type of the plate and the confidence level of each classification type of the tab.

[0017] In this way, by determining the confidence level of each classification type of electrode plate and electrode tab, the defect detection results of battery electrode sheets can be determined.

[0018] In some embodiments, the defect detection result characterizes the presence or absence of defects. Determining the defect detection result of the battery electrode by the confidence levels of each classification type of the electrode plate and each classification type of the electrode tab includes: determining that the defect detection result of the battery electrode is that there are no defects when both the confidence levels of the electrode plate classification type and the electrode tab classification type are greater than a preset threshold; and determining that the defect detection result of the battery electrode is that there are defects when at least one of the confidence levels of the electrode plate classification type and the electrode tab classification type is less than or equal to a preset threshold.

[0019] In this way, by setting a preset threshold for the confidence level when the classification type is normal, and by comparing the confidence levels of the plate and the tab when the classification type is normal with the preset threshold, the defect detection results of the battery electrode are determined, thereby enabling the judgment of whether there are defects in the battery electrode.

[0020] In some embodiments, after determining that the defect detection result of the battery electrode is that there is a defect, the method further includes: among the electrode plates and / or the tabs whose confidence level is less than or equal to a preset threshold and whose classification type is normal, determining the defect classification category with the highest confidence level other than normal, and obtaining the defect detection result of the battery electrode as the defect classification category.

[0021] In this way, by obtaining the defect classification category with the highest confidence level other than normal from the electrode plates and / or electrode tabs with a confidence level of less than or equal to a preset threshold, the defect detection results of the output battery electrode sheets include the defect category, which facilitates the differentiation and placement of different defect categories and the analysis of the causes of defects.

[0022] Secondly, embodiments of this application provide a defect detection device for a battery electrode sheet. The electrode sheet includes an electrode plate and a tab. The device includes: an acquisition module, configured to acquire a first image and a second image obtained by photographing the battery electrode sheet to be inspected, wherein the shooting parameters of the first image are preset parameters suitable for photographing the electrode plate area of ​​the battery electrode sheet; the shooting parameters of the second image are preset parameters suitable for photographing the tab area of ​​the battery electrode sheet; a first detection module, configured to perform image detection on the first image and the second image to obtain the electrode plate area and the tab area; and a second detection module, configured to perform defect detection on the electrode plate area and the tab area to obtain the defect detection result of the battery electrode sheet.

[0023] Thirdly, embodiments of this application provide an electronic device, the device comprising: a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the program to implement the steps in the above-described method.

[0024] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method.

[0025] This application proposes a defect detection method for battery electrodes based on deep learning and computer vision. By performing image detection on a first image and a second image, the electrode plate region and the electrode tab region are obtained; then, defect detection is performed on the electrode plate region and the electrode tab region to obtain the defect detection result of the battery electrode. Therefore, compared with manual inspection methods, this method not only improves the speed and accuracy of quality inspection but also reduces labor costs.

[0026] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of this application. Attached Figure Description

[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0028] Figure 1A A schematic diagram of the architecture of a defect detection system for battery electrode sheets provided in an embodiment of this application;

[0029] Figure 1B A schematic diagram of the architecture of another battery electrode defect detection system provided in an embodiment of this application;

[0030] Figure 1C A schematic flowchart illustrating a defect detection method for battery electrode sheets provided in an embodiment of this application;

[0031] Figure 2A A flowchart illustrating a method for determining the electrode plate region and the electrode tab region provided in an embodiment of this application;

[0032] Figure 2B and Figure 2C A flowchart illustrating another method for determining the electrode plate region and the electrode tab region provided in an embodiment of this application;

[0033] Figure 3A and Figure 3B A flowchart illustrating a method for determining the detection results of battery electrode defects, provided in an embodiment of this application;

[0034] Figure 4A A schematic diagram of a defect detection system for battery electrode sheets provided in an embodiment of this application;

[0035] Figure 4B A schematic diagram of a battery electrode image provided in an embodiment of this application;

[0036] Figure 4C This application provides a schematic diagram of a process for obtaining the electrode plate region and the electrode tab region in an embodiment of the present application.

[0037] Figure 5 A schematic diagram of a defect detection device for battery electrode sheets provided in an embodiment of this application;

[0038] Figure 6 A schematic diagram of a hardware entity of an electronic device provided in this application embodiment. Detailed Implementation

[0039] The technical solutions of this application will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0040] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0041] In the following description, the use of suffixes such as "module," "part," or "unit" to denote elements is solely for the purpose of illustration and has no specific meaning in itself. Therefore, "module," "part," or "unit" may be used interchangeably.

[0042] It should be noted that the terms "first, second, and third" used in the embodiments of this application are merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, and third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0043] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0044] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0045] Computer vision (CV) is a science that studies how to enable machines to "see." More specifically, it refers to machine vision, which uses cameras and computers to replace human eyes in tasks such as target recognition, tracking, and measurement, and further performs image processing to create images more suitable for human observation or transmission to instruments. As a scientific discipline, computer vision studies related theories and technologies, attempting to build artificial intelligence systems capable of extracting information from images or multidimensional data. Computer vision technologies typically include image processing, image recognition, image semantic understanding, image retrieval, optical character recognition (OCR), video processing, video semantic understanding, video content / behavior recognition, 3D object reconstruction, 3D graphics, virtual reality, augmented reality, simultaneous localization and mapping (SLAM), and common biometric recognition technologies such as facial recognition and fingerprint recognition.

[0046] Object detection algorithms aim to detect the presence and location of target objects in images or videos, and are usually output in the form of detection boxes.

[0047] Image semantic segmentation, simply put, is enabling computers to segment images based on their semantics. In the field of image processing, semantics refers to the content of an image, and segmentation means separating different objects in an image from the perspective of pixels, and recognizing pixels in the original image.

[0048] Image instance segmentation is essentially a combination of object detection and semantic segmentation. Compared to the bounding boxes of object detection, instance segmentation can be accurate down to the edges of objects; compared to semantic segmentation, instance segmentation can label different individuals of the same type of object in an image.

[0049] Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and learn-by-doing.

[0050] With the research and advancement of artificial intelligence (AI) technology, AI is being studied and applied in various fields, such as smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, autonomous driving, drones, robots, smart healthcare, and smart customer service. It is believed that with the development of technology, AI will be applied in more fields and play an increasingly important role.

[0051] The solutions provided in this invention relate to technologies such as machine learning and computer vision in artificial intelligence, and are specifically illustrated through the following embodiments:

[0052] Figure 1A A schematic diagram of an optional architecture of a battery electrode defect detection system 10 provided in an embodiment of this application is shown below. Figure 1AThe computer processing device 300 is connected to the image acquisition module 100 via a network 200. The network 200 can be a wide area network (WAN), a local area network (LAN), or a combination of both. The computer processing device 300 and the image acquisition module 100 can be physically separate or integrated. The image acquisition module 100 can send or store a first image and a second image, obtained by capturing images of the battery electrode to be inspected, to the computer processing device 300 via the network 200. After acquiring the first and second images, the computer processing device 300 performs image detection on the first and second images to obtain the electrode plate area and the electrode tab area; then, it performs defect detection on the electrode plate area and the electrode tab area to obtain the defect detection result of the battery electrode.

[0053] Figure 1B A schematic diagram of an optional architecture for another battery electrode defect detection system 10 provided in an embodiment of this application is shown below. Figure 1B The terminal device 500 is connected to the image acquisition module 100 via the network 200, and the terminal device 500 and the image acquisition module 100 are connected to the server 400 via the network 200. The image acquisition module 100 can send or store a first image and a second image of the battery electrode to be inspected, obtained by capturing images, to the server 400 via the network 200. After obtaining the first image and the second image, the server 400 performs image detection on the first image and the second image to obtain the electrode plate area and the electrode tab area; then, it performs defect detection on the electrode plate area and the electrode tab area to obtain the defect detection result of the battery electrode.

[0054] In some embodiments, server 400 may be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Image acquisition devices, electronic devices, and servers can be directly or indirectly connected via wired or wireless communication, which is not limited in this application embodiment.

[0055] This application provides a defect detection method for battery electrodes, applicable to electronic devices, such as... Figure 1C As shown, the method includes:

[0056] Step 102: Acquire a first image and a second image obtained by taking pictures of the battery electrode to be tested, wherein the shooting parameters of the first image are preset parameters suitable for shooting the plate area of ​​the battery electrode; the shooting parameters of the second image are preset parameters suitable for shooting the tab area of ​​the battery electrode.

[0057] Here, electronic devices can Figure 1A The computer processing device 300 in the middle, such as mobile phones, laptops, tablets, handheld internet devices, multimedia devices, streaming media devices, mobile internet devices, robots, etc.; can also be used for Figure 1B The server 400 in the system. The functions implemented by this method can be achieved by a processor in an electronic device calling program code. Of course, the program code can be stored in a computer storage medium. Therefore, the electronic device includes at least a processor and a storage medium. The processor can be used to perform defect detection of battery electrodes, and the memory can be used to store the data required and generated during the defect detection process of battery electrodes.

[0058] The battery may include, but is not limited to, lithium batteries, sodium batteries, hydrogen batteries, etc. The embodiments of this application do not limit the type of battery.

[0059] The first image and the second image can be images obtained by capturing the battery electrode to be tested under preset shooting parameters. The shooting parameters of the first image are suitable for capturing the plate area of ​​the battery electrode, and the shooting parameters of the second image are suitable for capturing the tab area of ​​the battery electrode. In some embodiments, the first image may only show the plate, and the second image may only show the tab; in other embodiments, both the plate and the tab may be shown in the first and second images; in some embodiments, both the plate and the tab may be shown in the first image, and only the tab may be shown in the second image; or the first image may only show the plate, and the second image may show both the plate and the tab. This application does not limit the objects displayed in the first and second images. Since the first and second images are used to subsequently acquire the areas where the plate and the tab are located, and to perform defect detection in these areas, the first image needs to clearly show the defects in the plate area; the second image needs to clearly show the defects in the tab area.

[0060] Since the materials on the surface of the electrode plate are usually positive and negative electrode pastes, mainly composed of oxides and non-metallic materials such as graphene, while the materials on the surface of the tab are generally copper foil and aluminum foil, mainly composed of metals, the different types of materials on their surfaces mean that, under the same shooting parameters, defects in the areas where the electrode plate and the tab are located cannot be clearly captured simultaneously. That is, if the defects in the area where the electrode plate is located are clear in the captured image, the defects in the area where the tab is located are not clear; and if the defects in the area where the tab is located are clear in the captured image, the defects in the area where the electrode plate is located are not clear. Moreover, even with image processing methods, it is impossible to process an image (e.g., an image where the electrode plate is clear) to clearly display the defects in another part (e.g., the area where the tab is located) for defect detection. Therefore, this application proposes to acquire a first image and a second image of the battery electrode to be tested under different shooting parameters for subsequent identification and defect detection of the electrode plate area and the tab area. In some embodiments, the light source for photographing the tab can be an arched light with an exposure time of 20 microseconds (μs); the light source for photographing the plate can be a high-angle linear light with an exposure time of 50 μs. The embodiments of this application do not limit the shooting parameters of the first image and the second image.

[0061] In some embodiments, the first image and the second image can be two-dimensional (2D) images or three-dimensional images. The 2D images can include red-green-blue (RGB) images and depth images captured by a monocular or multi-view camera, while the 3D images can include 3D images captured by a linear scanning camera. In some implementations, the first image and the second image can be images captured in real-time by an image acquisition device installed on the electronic device, such as a camera module. In other implementations, the first image and the second image can be images transmitted to the electronic device by other devices via instant messaging for defect detection of battery electrode sheets. In some implementations, the first image and the second image can also be images acquired by the electronic device in response to a task processing instruction, which are retrieved from the local photo album via a server. This embodiment of the application does not limit the scope of these embodiments.

[0062] In some embodiments, after acquiring the first image and the second image of the battery electrode to be detected in step 102, noise reduction processing can be performed on the first image and the second image to obtain a noise-reduced first image and a noise-reduced second image. This includes performing bilateral filtering and Gaussian filtering on the first image and the second image respectively to obtain the noise-reduced first image and the second image. For example, bilateral filtering and Gaussian filtering can be performed on the first image and the second image sequentially to obtain the noise-reduced first image and the second image; or, for another example, Gaussian filtering and bilateral filtering can be performed on the first image and the second image sequentially to obtain the noise-reduced first image and the second image. The filtering parameters of bilateral filtering and Gaussian filtering can be adjusted according to the appearance of the first image and the second image, shooting noise, etc., such as the shape and size of the filter kernel. By performing bilateral filtering and Gaussian filtering on the first image and the second image, imaging noise and interference from complex fine textures can be removed from the first image and the second image, thereby improving the accuracy of subsequent defect detection.

[0063] In some embodiments, the contrast between the first image and the second image can be enhanced to obtain a first image and a second image with enhanced contrast. For example, linear transformation, piecewise linear transformation, nonlinear transformation, histogram equalization, etc., can be used to enhance the contrast between the first image and the second image. For example, if the grayscale value range of the first image is (40, 150), the grayscale values ​​of the first image can be mapped to the interval (0, 255) to enhance the contrast of the first image. By enhancing the contrast between the first image and the second image, the grayscale difference between the first image and the second image can be increased, thereby helping the subsequent neural network to capture the defect features in the first image and the second image.

[0064] Step 104: Perform image detection on the first image and the second image to obtain the electrode plate region and the electrode tab region;

[0065] Here, the electrode plate region is the area including the electrode plate used for detecting electrode plate defects; the tab region is the area including the tab used for detecting tab defects.

[0066] In some embodiments, step 104 can be implemented by performing image detection on the plates and tabs in the first and second images using a neural network-based object detection algorithm to obtain the plate and tab regions. Examples include RetinaNet, YOLO, Fast R-CNN, and Faster R-CNN. In practice, a pre-trained network model of the object detection algorithm can be used for detection, and the bounding box output by the pre-trained network model represents the plate or tab region. In some embodiments, since the positions of the plate and tab regions are relatively fixed and their features are obvious, to improve the speed of the detection model, step 104 can employ a single-stage object detection algorithm to detect the plate and tab regions, such as RetinaNet, YOLO, and SSD. This allows for better application in production line processes and improves production efficiency. In some embodiments, since the production line conveyor belt moves very fast, the image detection speed can be further improved by changing the backbone network in the original single-stage target detection algorithm. For example, in the single-stage target detection algorithm RetinaNet, MobileNet can be used as the backbone to replace ResNet in the original network, making the detection speed of the electrode plate region and the electrode tab region faster.

[0067] In some embodiments, the pole plates and tabs in the first and second images can also be segmented using neural network-based image segmentation algorithms (such as semantic segmentation, instance segmentation, etc.) to obtain pole plate regions and tablature regions. Examples include Mask RCNN, PSPNet, DeepLabv3+, etc. In practice, a pre-trained network model of the image segmentation algorithm can be used for image segmentation, and the output connected components are the pole plate regions or tablature regions.

[0068] In some embodiments, the electrode region and electrode region in the first image and the second image can be extracted based on the mask images of the electrode plate and the electrode tab. In practice, different mask images can be pre-set for different battery models. Each model's mask image can include a masked region and a non-masked region. Taking the first image as an example, the masked region represents the position information of the electrode plate in the first image for that battery model, and the non-masked region represents the position information of the region other than the electrode plate in the first image. The first image is aligned with the mask image, and then the image region belonging to the masked region in the aligned first image is extracted as the electrode region, thereby achieving rapid acquisition of the electrode region.

[0069] The method for obtaining the electrode plate region and the electrode tab region in the embodiments of this application is not limited.

[0070] In some embodiments, step S104 can be implemented by performing image detection on the first image and the second image respectively to obtain the electrode plate region and the electrode tab region; alternatively, the first image and the second image can be stitched together to obtain a new image, and then image detection can be performed on the stitched image to obtain the electrode plate region and the electrode tab region. This can reduce the number of detections and improve detection efficiency. The embodiments of this application do not limit the method of detection for the first image and the second image.

[0071] When using target detection algorithms to detect the electrode plate region and the electrode tab region, such as Figure 2A As shown, the implementation of step 104 may include:

[0072] Step 1141: Using the target detection algorithm, perform image detection on the first image to obtain the plate detection box;

[0073] Here, the first image can be an image highlighting the electrode plate, obtained by photographing the battery electrode to be detected, clearly showing any defects in the electrode plate. In some embodiments, the electrode plate in the first image can be detected using a neural network-based object detection algorithm to obtain electrode plate detection boxes, such as RetinaNet, YOLO, Fast R-CNN, Faster R-CNN, etc. In practice, a first network model of a pre-trained object detection algorithm can be used for detection, and the output object detection box is the electrode plate detection box.

[0074] Step 1142: Using the target detection algorithm, perform image detection on the second image to obtain the tab detection box;

[0075] Here, the second image can be an image capturing the battery electrode to be inspected, highlighting the electrode tabs and clearly showing any defects in them. In some embodiments, the electrode tabs in the second image can be detected using a neural network-based object detection algorithm to obtain electrode tab detection boxes, such as RetinaNet, YOLO, Fast R-CNN, Faster R-CNN, etc. In practice, a pre-trained object detection algorithm's second network model can be used for detection, and the output object detection boxes are the electrode tab detection boxes.

[0076] It should be noted that the detection of the first image and the second image can be performed separately, or it can be performed simultaneously by combining the first image and the second image (in which case the first network model and the second network model are the same network model). When the first image and the second image are combined and detected simultaneously, the number of electrode detection boxes and electrode tab detection boxes obtained are at least one each.

[0077] Step 1143: Cut the electrode plate detection frame and the electrode tab detection frame respectively to obtain the electrode plate region and the electrode tab region.

[0078] Here, when the first and second images are stitched together and detected simultaneously, since at least one plate detection box and one tab detection box are obtained, the plate detection boxes and tab detection boxes used for cropping can be determined based on the relative positional relationship between the first and second images in the stitched image. For example, if the first image is to the left of the second image, then the plate detection box on the left is the plate detection box to be cropped, and the tab detection box on the right is the tab detection box to be cropped.

[0079] In this embodiment, the electrode plate detection frame and the electrode tab detection frame are cut out for subsequent defect detection, which can reduce the interference of the background in the first and second images on the defect detection results.

[0080] In some embodiments, such as Figure 2B As shown, the implementation of step 1143 may include:

[0081] Step 1431: Expand the electrode plate detection frame and the electrode tab detection frame outward to obtain the electrode plate area to be processed and the electrode tab area to be processed;

[0082] Here, since defects on the electrode plate and tab may exist at their edges, to improve the accuracy of defect detection, in step 1431, the electrode plate detection frame and the tab detection frame can be expanded outwards respectively to obtain the electrode plate area to be processed and the tab area to be processed, so that the electrode plate area to be processed and the tab area to be processed can include defects at the edges of the electrode plate and the tab. In some embodiments, the electrode plate detection frame and the tab detection frame can be expanded outwards by 5% to 10%. For example, if the tab detection frame has 100*100 pixels, expanding it by 10% results in 10 pixels. In practice, the center of the tab detection frame can be used as the origin, and 10 pixels can be added up, down, left, and right to obtain the tab area to be processed. The outward expansion size is not limited in this embodiment.

[0083] Step 1432: Cut the electrode plate area to be processed and the electrode tab area to be processed respectively to obtain the electrode plate area and the electrode tab area.

[0084] Here, step 1432 can be found in step 1143.

[0085] In this embodiment, by expanding the electrode plate detection frame and the electrode tab detection frame outward, the electrode plate to be processed area and the electrode tab to be processed area are obtained respectively, so that the electrode plate to be processed area and the electrode tab to be processed area can include defects at the edge of the electrode plate and the electrode tab, thereby improving the accuracy of subsequent defect detection.

[0086] In some embodiments, such as Figure 2C As shown, the implementation of step 104 may include:

[0087] Step 1241: Stitch the first image and the second image together to obtain a battery electrode image;

[0088] Here, step 1241 involves stitching the first image and the second image together into a single image, namely the battery electrode image, which is then used for image detection in the subsequent step 1242, thereby achieving end-to-end defect detection. Since the channel directions of the first and second images cannot be aligned during stitching (otherwise, the number of channels would increase, preventing the detection network from performing image detection and processing), the number of channels in the first and second images should be the same. For example, if the first image has 3 channels, the second image should also have 3 channels. If the number of channels in the first and second images differs, a dimensionality transformation method can be used to convert them into two images with the same number of channels.

[0089] If the first and second images have the same number of channels, then in order to stitch them into a single image, the first and second images should have the same size in at least one of the width and height dimensions. If the sizes are the same in both width and height dimensions, the battery electrode image can be obtained by stitching along either the width or height dimension. If the sizes are the same in only one of the width and height dimensions, the battery electrode image can be obtained by stitching along different dimensions; for example, if the sizes are the same in the width dimension, then the battery electrode image can be obtained by stitching along the height dimension.

[0090] The following example illustrates this: both the first and second images have 3 channels, the first image has a pixel size of 4000 (width dimension) * 8000 (height dimension), and the second image has a pixel size of 4000 (width dimension) * 6000 (height dimension). Since the first and second images have the same pixel size of 4000 in the width dimension, they can be stitched together along the height dimension to obtain a battery electrode image with a pixel size of 4000 * 14000.

[0091] When the first image and the second image do not have the same size in either the width or height dimension, a dimensional transformation method can be used to make the first image and the second image have the same size in at least one of the width and height dimensions. The following example illustrates this: the first image has a pixel size of 4000 (width dimension) * 6000 (height dimension), and the second image has a pixel size of 2000 (width dimension) * 4000 (height dimension). In practice, a dimensional transformation method can be used to convert the pixel size of the second image's width dimension to 4000, resulting in a second image pixel size of 4000 (width dimension) * 8000 (height dimension), thus making the pixel size of the first image and the second image the same in the width dimension, both being 4000.

[0092] Step 1242: Using the target detection algorithm, simultaneously perform electrode plate detection and electrode tab detection on the first image and the second image in the battery electrode image to obtain at least one electrode plate detection box and at least one electrode tab detection box.

[0093] Here, the number of electrode plate detection boxes and electrode tab detection boxes depends on whether the first image and the second image include electrodes and tabs. When both the first image and the second image include electrodes and tabs, the number of electrode plate detection boxes is 2 and the number of electrode tab detection boxes is 2; when the first image includes electrodes and tabs and the second image includes electrodes, the number of electrode plate detection boxes is 1 and the number of electrode tab detection boxes is 2; when the first image includes electrodes and the second image includes electrodes, the number of electrode plate detection boxes is 1 and the number of electrode tab detection boxes is 1.

[0094] Step 1243: In the at least one electrode plate detection frame and the at least one electrode tab detection frame, the electrode plate region and the electrode tab region are determined based on the relative positional relationship between the first image and the second image in the battery electrode image.

[0095] Here, the relative positional relationship between the first image and the second image can be: the first image is located to the left of the second image; the first image is located to the right of the second image; the first image is located above the second image; the first image is located below the second image, etc. The embodiments of this application do not limit the relative positional relationship between the first image and the second image.

[0096] In some embodiments, step 1243 may include: when there are two electrode plate detection frames or two electrode tab detection frames, by obtaining the coordinates of the center points of the two electrode plate detection frames or two electrode tab detection frames, comparing the positional relationship of the coordinates of the two center points, and determining which electrode plate detection frame belongs to the first image and which electrode tab detection frame belongs to the second image based on the relative positional relationship between the first image and the second image in the battery electrode image, thereby determining the electrode plate region and the electrode tab region. For example, if the first image is located to the left of the second image, and the coordinates of the center point of the first electrode plate detection frame are located to the left of the coordinates of the center point of the second electrode plate detection frame, then the first electrode plate detection frame is used to determine the electrode plate region.

[0097] In some embodiments, after the implementation of step 104, the process further includes: aligning the electrode plate region and the electrode tab region.

[0098] During implementation, a target image can be preset, which can be an image of the electrode plate area or electrode tab area that does not form an angle with the target direction. If the obtained electrode plate area or electrode tab area is not in the correct position, i.e., it forms an angle with the target direction, the electrode plate area or electrode tab area can be aligned with the target image. Then, based on the aligned electrode plate area and electrode tab area, the defect detection result of the battery electrode is determined, thereby improving the accuracy of subsequent defect detection.

[0099] In this embodiment of the application, a battery electrode image is obtained by stitching together the first image and the second image. Then, image detection is performed on the battery electrode image to obtain the plate area and the tab area. Finally, defect detection is performed on the plate area and the tab area to achieve end-to-end detection of battery electrode defects.

[0100] Step 106: Perform defect detection on the electrode plate area and the electrode tab area to obtain the defect detection results of the battery electrode.

[0101] Here, the defect detection result indicates whether a defect exists or not. If the defect detection result indicates that a defect exists, the defect detection result may also include the type of defect, such as breakage, exposed material, wrinkles, material loss, foreign matter, color difference, cracks, etc. In some embodiments, if the defect detection result indicates that a defect exists, the defect detection result may not include the type of defect, but only output that the defect exists.

[0102] In some embodiments, step 106 can be implemented by using a neural network to detect defects in the electrode plate region and the tab region. In practice, the electrode plate region and the tab region can be stitched together into an image, which is then fed into the neural network for defect detection to obtain the defect detection result of the battery electrode.

[0103] In some embodiments, step 106 can also be implemented by using two neural networks to detect defects in the plate region and the tab region respectively. In this implementation, a third network model can be used to detect defects in the plate region, and a fourth network model can be used to detect defects in the tab region. The defect detection results of the two neural networks are then combined to obtain the defect detection results of the battery electrode. Since the plate region is larger and has more defect categories, while the tab region is smaller and has fewer defect categories, the accuracy of the third network model can be relatively higher than that of the fourth network model. Therefore, an appropriate network model can be used reasonably to accurately and quickly detect defects in the electrode.

[0104] In some embodiments, step 106 can be implemented by using an image classification network to detect the electrode plate region and the tab region to obtain the defect detection result of the battery electrode. Examples include VGG Net, ResNet, ResNeXt, and SE-Net. In practice, a pre-trained image classification network can be used for detection, and the category with the highest confidence output by the pre-trained image classification network is the defect detection result of the battery electrode. Alternatively, a neural network-based object detection algorithm can be used to detect the electrode plate region and the tab region to obtain the defect detection result of the battery electrode. Examples include RetinaNet, YOLO, Fast R-CNN, and Faster R-CNN. In practice, a pre-trained network model of the object detection algorithm can be used for detection, and the target detection box output by the pre-trained network model is the location of the defect in the battery electrode. Alternatively, a neural network-based image segmentation algorithm (semantic segmentation, instance segmentation, etc.) can be used to segment the defects in the first and second images to obtain the defect detection result of the battery electrode. Examples include Mask. When implementing algorithms such as RCNN, PSPNet, and DeepLabv3+, a pre-trained network model of an image segmentation algorithm can be used for image segmentation. The target connected region in the potential defect region obtained by the pre-trained network model of the image segmentation algorithm is the location of the defect in the battery electrode. The target connected region is the connected region in the potential defect region whose area is greater than a threshold.

[0105] In some embodiments, the network model described above can be trained using a training image set, enabling the network model to identify defects in the plate region and the tab region.

[0106] This application proposes a defect detection method for battery electrodes based on deep learning and computer vision. By performing image detection on a first image and a second image, the electrode plate region and the electrode tab region are obtained; then, defect detection is performed on the electrode plate region and the electrode tab region to obtain the defect detection result of the battery electrode. Therefore, compared with manual inspection methods, this method not only improves the speed and accuracy of quality inspection but also reduces labor costs.

[0107] In some embodiments, the implementation of step 106 may include:

[0108] The electrode plate area and the tab area are classified using classification algorithms respectively to obtain the defect detection results of the battery electrode.

[0109] The first classification algorithm is used to classify the plate area, and the second classification algorithm is used to classify the tab area, ultimately obtaining the defect detection results of the battery electrode.

[0110] The reason for using two classification algorithms to determine the defect detection results of the battery electrode is as follows:

[0111] On the one hand, for the actual production process on the production line, it is only necessary to determine that there is a defect in the product, and then consider the product as a defective product and put it into the defective product area. It is not necessary to know the specific location of the defect. In addition, compared with object detection algorithms and image segmentation algorithms, the defect labeling in the early stage of classification detection algorithms takes less time and the application process is faster.

[0112] On the other hand, the size difference between the electrode plate and the electrode tab in the electrode sheet is too large (usually the electrode tab is about 100*100 pixels and the electrode plate is about 4000*6000 pixels), and there are many types of defects. If the same classification algorithm is used to detect defects in the electrode plate area and the electrode tab area, the defects in the electrode tab area cannot be accurately determined.

[0113] Therefore, in step 106, two classification algorithms are used to detect defects in the electrode plate area and the electrode tab area respectively. This can improve the accuracy of defect detection and be better applied to the fast-paced production process of the production line, thereby improving production efficiency.

[0114] During implementation, a pre-trained fifth network model with a classification algorithm can be used to detect defects in the electrode area and obtain the defect detection results of the electrode area.

[0115] In some embodiments, the fifth network model may be trained using a first training image set, wherein the training images in the first training image set include labeled data of defects. The labeled data of defects may be labeled data including defect categories. In this example, by training the fifth network model using a first training image set with labeled data of defects, the fifth network model is able to learn the ability to identify defect types in the electrode region.

[0116] Similarly, a pre-trained sixth network model with a classification algorithm can be used to detect the tab region and obtain the defect detection results of the tab region.

[0117] In some embodiments, the sixth network model may be trained using a second training image set, wherein the training images in the second training image set include labeled data of defects. The labeled data of defects may include labeled data that includes defect categories. In this example, by training the sixth network model using a second training image set with labeled data of defects, the sixth network model is able to learn the ability to identify defect types in the tab region.

[0118] When training the fifth and sixth network models, positive and negative samples can be used for training. Positive samples can be images containing a certain type of defect, such as damage, while negative samples can be images not containing that type of defect. Since the size and angle of the same defect type can vary, to improve the detection rate of the fifth and sixth network models for defects of different sizes and angles, the positive and negative samples can be scaled by different ratios or rotated by different angles before being fed into the fifth and sixth network models for training.

[0119] Furthermore, since the number of sample images of a certain type of defect in the battery electrode is relatively small, when training the fifth and sixth network models, the sample images of this type can be resampled. For example, methods such as scaling, flipping, adjusting contrast and brightness can be used to increase the number of sample images, thereby improving the detection rate of this type of defect.

[0120] In this embodiment, by using classification algorithms to detect defects in the plate area and the tab area respectively, the defect detection results of the battery electrode are obtained. This not only makes the defect detection speed fast and can be applied to high-speed production lines, but also improves the accuracy of defect detection.

[0121] In some embodiments, the classification types of electrode plates include multiple types from: damaged, exposed, wrinkled, material loss, foreign matter, color difference, crack, and normal; the classification types of electrode tabs include multiple types from: damaged, misaligned, and normal, such as... Figure 3A As shown, the implementation of step 106 includes:

[0122] Step 1061: Classify the electrode plate region and the electrode tab region using a classification algorithm to obtain the confidence level of each classification type of the electrode plate and the confidence level of each classification type of the electrode tab;

[0123] Here, since step 1061 requires obtaining the confidence levels for each category of the electrode plate (i.e., damage, exposed white, wrinkles, material loss, foreign matter, color difference, cracks, and normal) and each category of the electrode tab (i.e., damage, misalignment, and normal), the Sigmoid function can be used to output the confidence level for each category. The confidence level is the score obtained after applying the Sigmoid function, representing the probability of each defect type. A higher confidence level indicates a greater probability of belonging to that defect type. For example, a confidence level of 0.9 indicates a higher probability of belonging to a certain defect type than a confidence level of 0.8.

[0124] Step 1062: Determine the defect detection result of the battery electrode based on the confidence level of each classification type of the electrode plate and the confidence level of each classification type of the electrode tab.

[0125] In this way, by determining the confidence level of each classification type, the defect detection results of the battery electrode can be determined.

[0126] In some embodiments, defect detection results characterize the presence or absence of defects, such as Figure 3B As shown, the implementation of step 1062 may include:

[0127] Step 1062a: If the confidence level of the classification type of the electrode plate being normal and the confidence level of the classification type of the electrode tab being normal are both greater than a preset threshold, the defect detection result of the battery electrode is determined to be that there is no defect.

[0128] Here, the preset threshold can be set using empirical values ​​or historical data; for example, the preset threshold can be 0.5. Since step 1061 outputs the confidence level for each classification type, the confidence levels for the plate classification type being "normal" and the tab classification type being "normal" can be obtained. The following explanation uses a preset threshold of 0.5 as an example. If the confidence level for the plate classification type being "normal" is 0.6 and the confidence level for the tab classification type being "normal" is 0.7, and both confidence levels are greater than 0.5, then the defect detection result for the battery electrode is that there are no defects.

[0129] Step 1062b: If at least one of the confidence levels of the electrode plate classification type being normal and the confidence level of the electrode tab classification type being normal is less than or equal to a preset threshold, the defect detection result of the battery electrode is determined to be defective.

[0130] Here, the implementation of determining the defect detection result of the battery electrode in step 1062b as having a defect can include the following situations:

[0131] The first type is where the confidence level for the plate classification type to be normal is greater than the preset threshold, and the confidence level for the ear classification type to be normal is less than or equal to the preset threshold.

[0132] The second type is where the confidence level for the plate classification type is normal is less than or equal to the preset threshold, and the confidence level for the ear classification type is normal is greater than the preset threshold.

[0133] The third type is where the classification type of the electrode plate is normal with a confidence level less than or equal to a preset threshold, and the classification type of the electrode ear is normal with a confidence level less than or equal to a preset threshold.

[0134] The following explanation continues using a preset threshold of 0.5 as an example. If the confidence level for the classification type of the electrode plate is normal is 0.3 and the confidence level for the classification type of the tab is normal is 0.7, then the defect detection result of the battery electrode is that there is a defect.

[0135] In this embodiment, a preset threshold for the confidence level when the classification type is normal is set, and the defect detection result of the battery electrode is determined by comparing the confidence level of the electrode plate when the classification type is normal and the electrode tab when the classification type is normal with the preset threshold, thereby realizing the judgment of whether there is a defect in the battery electrode.

[0136] In some embodiments, such as Figure 3B As shown, after confirming that the defect detection result of the battery electrode sheet indicates the presence of a defect, the following steps are also included:

[0137] Step 1062c: Among the electrode plates and / or electrode tabs whose confidence level is less than or equal to a preset threshold and classified as normal, determine the defect classification category with the highest confidence level other than normal, and obtain the defect detection result of the battery electrode as the defect classification category.

[0138] Here, the classification types of the electrode plates and the electrode tabs, excluding normal, include electrode plate damage, exposed material, wrinkles, material loss, foreign matter, color difference, and cracks, and electrode tab damage and misalignment. That is, in step 1062b, among the electrode plates and / or electrode tabs whose confidence level for the normal classification type is less than or equal to a preset threshold, the defect classification type with the highest confidence level other than normal is determined as the defect detection result of the battery electrode.

[0139] The following explanation continues using a preset threshold of 0.5, a confidence level of 0.3 for the normal plate classification type, and a confidence level of 0.7 for the normal tab classification type. Since the confidence level for the normal plate classification type is less than 0.5, the defect detection result for the battery electrode is that a defect exists. Among the plate classification types other than normal (damaged (confidence 0.1), exposed white (confidence 0.7), wrinkled (confidence 0.1), material loss (confidence 0.2), foreign matter (confidence 0.3), color difference (confidence 0.2), crack (confidence 0.1) and tab classification types (damaged (confidence 0.2), misaligned (confidence 0.1), the defect classification type with the highest confidence is exposed white. Therefore, the defect detection result for the battery electrode is exposed white.

[0140] In some embodiments, a threshold can be set to output all defect classification types greater than the threshold as the defect detection result of the battery electrode. This application embodiment does not limit the output format of the defect detection result of the battery electrode.

[0141] In this embodiment of the application, by determining the defect classification category with the highest confidence level other than normal among the electrode plates and / or tabs with a confidence level of less than or equal to a preset threshold, the defect detection results of the output battery electrode sheet include the defect category, which facilitates the differentiation, placement and analysis of the causes of defects for different defect categories.

[0142] This application also provides a defect detection method for battery electrodes, applied to a defect detection system for battery electrodes, such as... Figure 4A As shown, the system includes: an electrode image input module 401, an electrode tab positioning module 402, an electrode defect detection module 403, an electrode tab defect detection module 404, and a result display output module 405.

[0143] The electrode image input module 401 includes an industrial camera and a light source, which can be mounted at a fixed position on the electrode production line. Each electrode on the production line can be illuminated by two light sources, one for emphasizing the electrode plate area and the other for capturing images of the tab area. Figure 4B As shown, the final image is an image 406 (i.e., the battery electrode image) composed of two images taken by different light sources, stitched together from left to right. The left side is the light source image 4061 (i.e., the first image) that emphasizes the electrode area, and the right side is the light source image 4062 (i.e., the second image) that emphasizes the electrode area. This stitched image 406 is then input into the edge node algorithm module for processing.

[0144] The edge node algorithm module includes an electrode plate and tab localization module 402, an electrode plate defect detection module 403, and an electrode sheet defect detection module 404. Since the stitched image contains significant background interference, directly detecting defects in the electrode sheet would reduce accuracy. Furthermore, the electrode sheet consists of a lower electrode plate and an upper tab, and the types of defects in these two parts differ, making it difficult to distinguish specific defect types through direct defect identification. Therefore, this application proposes an algorithm that first detects the electrode plate and tab regions, and then performs defect detection on the left electrode plate and right tab respectively, including the following steps:

[0145] Part 1: Plate and tab detection.

[0146] Since the positions of the pole plates and tabs are relatively fixed and their features are obvious, in order to improve the speed of the detection model, the single-stage target detection method RetinaNet, with MobileNet as the backbone, is used to detect the pole plate and tab regions. Figure 4C Two electrode plates 407 and two electrode tabs 408 were detected.

[0147] Since the image input module receives an image composed of two stitched images—the left image emphasizing the light source region of the electrode plate and the right image emphasizing the light source region of the electrode tabs—RetinaNet will detect two electrode plates and tabs. The left electrode plate bounding box and the right tab bounding box are selected as the identified electrode plate and tab regions. During the cropping process, the tab detection box is expanded by 10 pixels and the electrode plate detection box by 20 pixels before cropping. The result is as follows: Figure 4C The electrode plate region diagram 409 (i.e., the electrode plate region) and the electrode tab region diagram 410 (i.e., the electrode tab region) shown are used to identify corresponding electrode plate defects and electrode tab defects, respectively.

[0148] Part Two: Defect Detection.

[0149] Due to the larger area and greater variety of defects in the electrode plate region, ResNet50 is used as the electrode plate defect classification model. Due to the smaller area and fewer defect categories in the electrode sheet region, ResNet18 is used as the electrode sheet defect classification model. Electrode plate defects include six categories: breakage, exposed material, wrinkles, material loss, foreign matter, color difference, and cracks. Electrode tab defects include two categories: breakage and misalignment. In addition to these defect categories, there are also normal images; therefore, the electrode plate defect classification model performs seven classifications, while the electrode sheet defect classification model performs three.

[0150] Since the number of defective images varies, in order to improve the classification model's accuracy in recognizing these categories, images of defective categories with fewer defects are resampled. Also, since some images have multiple defects, the sigmoid function is used instead of the softmax function for each type, meaning that the score for each type is between 0 and 1.

[0151] Part Three: Defect Results.

[0152] The display and control module 405 is used to display the results of the edge node algorithm module. If the normal class scores of the electrode plate classification model and the electrode sheet classification model are both greater than 0.5 (i.e., the preset threshold), the electrode sheet is determined to be a good product. Otherwise, there is a defect. The defect type is determined to be the defect with the highest score in the defect category. Then, the corresponding robotic arm can be controlled by the process production line controller (PLC) to grab the defective product (No Good, NG) to the corresponding NG product area and directly remove the unqualified product from the production line.

[0153] It should be noted that the descriptions of the various embodiments above tend to emphasize the differences between them, while their similarities or commonalities can be referred to each other. Those skilled in the art will understand that in the methods described above in the specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The execution order of each step should be determined by its function and possible internal logic.

[0154] Based on the foregoing embodiments, this application provides a defect detection device for battery electrode sheets. The device includes various modules, sub-modules, units, and sub-units, all of which can be implemented by electronic devices; of course, they can also be implemented by specific logic circuits. In the implementation process, the processor can be a central processing unit (CPU), microprocessor (MPU), digital signal processor (DSP), or field programmable gate array (FPGA), etc.

[0155] Figure 5 This is a schematic diagram of the structure of a defect detection device for battery electrode sheets provided in an embodiment of this application, as shown below. Figure 5 As shown, the defect detection device 500 includes an acquisition module 510, a first detection module 520, and a second detection module 530, wherein:

[0156] The acquisition module 510 is used to acquire a first image and a second image obtained by taking pictures of the battery electrode to be tested, wherein the shooting parameters of the first image are preset parameters suitable for shooting the electrode plate area of ​​the battery electrode; and the shooting parameters of the second image are preset parameters suitable for shooting the tab area of ​​the battery electrode.

[0157] The first detection module 520 is used to perform image detection on the first image and the second image to obtain the electrode plate region and the electrode tab region.

[0158] The second detection module 530 is used to perform defect detection on the electrode plate area and the electrode tab area to obtain the defect detection results of the battery electrode.

[0159] In some embodiments, the first detection module 520 is further configured to perform image detection on the first image and the second image using a target detection algorithm to obtain the electrode plate region and the electrode tab region.

[0160] In some embodiments, the first detection module 520 includes:

[0161] The first detection submodule is used to perform image detection on the first image using the target detection algorithm to obtain the plate detection box;

[0162] The second detection submodule is used to perform image detection on the second image using the target detection algorithm to obtain the electrode detection box;

[0163] The trimming submodule is used to trim the electrode plate detection frame and the electrode tab detection frame respectively to obtain the electrode plate region and the electrode tab region.

[0164] In some embodiments, the first detection module 520 includes: a stitching submodule, used to stitch the first image and the second image to obtain a battery electrode image; a third detection submodule, used to use the target detection algorithm to simultaneously perform electrode plate detection and electrode tab detection on the first image and the second image in the battery electrode image to obtain at least one electrode plate detection frame and at least one electrode tab detection frame; and a first determination submodule, used to determine the electrode plate region and the electrode tab region in the at least one electrode plate detection frame and the at least one electrode tab detection frame based on the relative positional relationship between the first image and the second image in the battery electrode image.

[0165] In some embodiments, the cutting submodule includes: an expansion unit for expanding the electrode plate detection frame and the electrode tab detection frame to obtain an electrode plate area to be processed and an electrode tab area to be processed; and a cutting unit for cutting the electrode plate area to be processed and the electrode tab area to be processed to obtain the electrode plate area and the electrode tab area.

[0166] In some embodiments, the second detection module 530 is further configured to classify the electrode plate region and the electrode tab region respectively using a classification algorithm to obtain the defect detection result of the battery electrode.

[0167] In some embodiments, the electrode plate is classified into multiple types including: damaged, exposed white, wrinkled, missing material, foreign matter, color difference, crack, and normal; the tab is classified into multiple types including: damaged, misaligned, and normal.

[0168] The second detection module 530 includes: a classification submodule, used to classify the electrode plate region and the electrode tab region using a classification algorithm to obtain the confidence level of each classification type of the electrode plate and the confidence level of each classification type of the electrode tab; and a second determination submodule, used to determine the defect detection result of the battery electrode based on the confidence level of each classification type of the electrode plate and the confidence level of each classification type of the electrode tab.

[0169] In some embodiments, the defect detection result represents the presence or absence of defects. The second determining submodule includes: a first determining unit, configured to determine that the defect detection result of the battery electrode is that there are no defects when both the confidence level of the electrode plate's classification type being normal and the confidence level of the electrode tab's classification type being normal are greater than a preset threshold; and a second determining unit, configured to determine that the defect detection result of the battery electrode is that there are defects when at least one of the confidence levels of the electrode plate's classification type being normal and the confidence level of the electrode tab's classification type being normal is less than or equal to a preset threshold.

[0170] In some embodiments, after determining that the defect detection result of the battery electrode is that there is a defect, the second determining unit is further configured to determine the defect classification category with the highest confidence other than normal among the electrode plates and / or the electrode tabs whose confidence level of the classification type is normal is less than or equal to a preset threshold, so as to obtain the defect detection result of the battery electrode as the defect classification category.

[0171] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0172] It should be noted that, in the embodiments of this application, if the above-mentioned defect detection method for battery electrode sheets is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware, software, or firmware, or any combination of hardware, software, and firmware.

[0173] Correspondingly, this application provides an electronic device including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the steps in the defect detection method for battery electrodes provided in the above embodiments.

[0174] Correspondingly, embodiments of this application provide a readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps in the above-described method for detecting defects in battery electrodes.

[0175] It should be noted that the descriptions of the storage media and platform embodiments above are similar to those of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage media and platform embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0176] It should be noted that, Figure 6 A schematic diagram of a hardware entity of an electronic device provided in an embodiment of this application, such as... Figure 6 As shown, the hardware entity of the electronic device 600 includes a processor 601, a communication interface 602, and a memory 603, wherein the processor 601 typically controls the overall operation of the electronic device 600. The communication interface 602 enables the electronic device 600 to communicate with other platforms, electronic devices, or servers via a network.

[0177] The memory 603 is configured to store instructions and applications executable by the processor 601, and can also cache data to be processed or already processed (e.g., image data, audio data, voice communication data and video communication data) in the processor 601 and various modules in the electronic device 600. It can be implemented by FLASH or random access memory (RAM).

[0178] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0179] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0180] Furthermore, in the various embodiments of this application, all functional units can be integrated into one processing module, or each unit can be a separate unit, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units. Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0181] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0182] The features disclosed in the several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0183] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0184] If the technical solution of this application involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the user through pop-up information or by asking the user to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

[0185] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method of defect detection of a battery electrode sheet, characterized by, The pole piece includes a pole plate and a tab, and the method includes: obtaining a first image and a second image obtained by photographing a battery pole piece to be detected, wherein the photographing parameter of the first image is a preset parameter suitable for photographing the pole plate region of the battery pole piece; and the photographing parameter of the second image is a preset parameter suitable for photographing the tab region of the battery pole piece; image detection is performed on the first image and the second image to obtain a pole plate region and a tab region; defect detection is performed on the pole plate region and the tab region to obtain a defect detection result of the battery pole piece; the defect detection on the pole plate region and the tab region to obtain the defect detection result of the battery pole piece includes: different classification algorithms are used to classify the pole plate region and the tab region respectively to obtain the confidence of each classification type of the pole plate and the confidence of each classification type of the tab; the classification types of the pole plate include multiple types of damage, white exposure, wrinkles, material loss, foreign matter, color difference, cracks and normal; and the classification types of the tab include multiple types of damage, misplacement and normal; based on the confidence of each classification type of the pole plate and the confidence of each classification type of the tab, the defect detection result of the battery pole piece is determined.

2. The method of claim 1, wherein, the image detection on the first image and the second image to obtain a pole plate region and a tab region includes: a target detection algorithm is used to perform image detection on the first image and the second image to obtain a pole plate region and a tab region.

3. The method of claim 2, wherein, the image detection on the first image and the second image to obtain a pole plate region and a tab region using a target detection algorithm includes: the target detection algorithm is used to perform image detection on the first image to obtain a pole plate detection frame; the target detection algorithm is used to perform image detection on the second image to obtain a tab detection frame; the pole plate detection frame and the tab detection frame are respectively cropped to obtain the pole plate region and the tab region.

4. The method of claim 2, wherein, the image detection on the first image and the second image to obtain a pole plate region and a tab region using a target detection algorithm includes: the first image and the second image are spliced to obtain a battery pole piece image; the target detection algorithm is used to simultaneously perform pole plate detection and tab detection on the first image and the second image in the battery pole piece image to obtain at least one pole plate detection frame and at least one tab detection frame; based on the relative position relationship between the first image and the second image in the battery pole piece image, the pole plate region and the tab region are determined in the at least one pole plate detection frame and the at least one tab detection frame.

5. The method of claim 3, wherein, the pole plate detection frame and the tab detection frame are respectively cropped to obtain the pole plate region and the tab region. the pole plate detection frame and the tab detection frame are respectively cropped to obtain the pole plate region and the tab region. the pole plate detection frame and the tab detection frame are respectively cropped to obtain the pole plate region and the tab region.

6. The method of claim 1, wherein, The defect detection result represents presence of a defect and absence of a defect, confidence of each classification type of the plate, and confidence of each classification type of the lug, and the defect detection result of the battery pole piece is determined by: In a case where the confidence of the classification type of the plate and the confidence of the classification type of the lug are both greater than a preset threshold, the defect detection result of the battery pole piece is determined as absence of a defect; In a case where at least one of the confidence of the classification type of the plate and the confidence of the classification type of the lug is less than or equal to the preset threshold, the defect detection result of the battery pole piece is determined as presence of a defect.

7. The method of claim 6, wherein, After determining that the defect detection result of the battery pole piece is presence of a defect, the method further comprises: In the plate and / or the lug with the confidence of the classification type being less than or equal to the preset threshold, a defect classification category with the highest confidence except for the normal classification type is determined, and the defect detection result of the battery pole piece is obtained as the defect classification category.

8. A battery electrode sheet defect detection apparatus characterized by comprising: The pole piece includes a plate and a lug, and the device comprises: An acquisition module is configured to acquire a first image and a second image obtained by photographing a battery pole piece to be detected, wherein a photographing parameter of the first image is a preset parameter suitable for photographing a plate region of the battery pole piece, and a photographing parameter of the second image is a preset parameter suitable for photographing a lug region of the battery pole piece; A first detection module is configured to perform image detection on the first image and the second image to obtain a plate region and a lug region; A second detection module is configured to perform defect detection on the plate region and the lug region to obtain a defect detection result of the battery pole piece; The second detection module is further configured to classify the plate region and the lug region by using different classification algorithms respectively to obtain confidence of each classification type of the plate and confidence of each classification type of the lug, wherein the classification types of the plate include multiple types of damage, white exposure, wrinkle, material loss, foreign matter, color difference, crack, and normal, and the classification types of the lug include multiple types of damage, misposition, and normal; and the defect detection result of the battery pole piece is determined based on the confidence of each classification type of the plate and the confidence of each classification type of the lug.

9. An electronic device, comprising: A memory and a processor are included, the memory stores a computer program capable of running on the processor, and the processor implements the steps in the method of any one of claims 1 to 7 when executing the program.

10. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program implements the steps in the method of any one of claims 1 to 7 when executed by a processor.

Images