WO2020042741A1 - 电池检测方法及装置 - Google Patents

电池检测方法及装置 Download PDF

Info

Publication number
WO2020042741A1
WO2020042741A1 PCT/CN2019/093383 CN2019093383W WO2020042741A1 WO 2020042741 A1 WO2020042741 A1 WO 2020042741A1 CN 2019093383 W CN2019093383 W CN 2019093383W WO 2020042741 A1 WO2020042741 A1 WO 2020042741A1
Authority
WO
WIPO (PCT)
Prior art keywords
defect
detection model
defect detection
battery
picture
Prior art date
Application number
PCT/CN2019/093383
Other languages
English (en)
French (fr)
Inventor
文亚伟
冷家冰
刘明浩
肖慧慧
郭江亮
李旭
Original Assignee
北京百度网讯科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京百度网讯科技有限公司 filed Critical 北京百度网讯科技有限公司
Priority to EP19853678.1A priority Critical patent/EP3683572B1/en
Priority to US16/650,279 priority patent/US11158044B2/en
Priority to KR1020207036130A priority patent/KR102494946B1/ko
Priority to JP2020519278A priority patent/JP7018135B2/ja
Publication of WO2020042741A1 publication Critical patent/WO2020042741A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
    • H01M10/00—Secondary cells; Manufacture thereof
    • H01M10/42—Methods or arrangements for servicing or maintenance of secondary cells or secondary half-cells
    • H01M10/4285—Testing apparatus
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G06T7/001—Industrial image inspection using an image reference approach
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
    • H01M6/00—Primary cells; Manufacture thereof
    • H01M6/50—Methods or arrangements for servicing or maintenance, e.g. for maintaining operating temperature
    • H01M6/5083—Testing apparatus
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8806—Specially adapted optical and illumination features
    • G01N2021/8809—Adjustment for highlighting flaws
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8883—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges involving the calculation of gauges, generating models
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/10—Energy storage using batteries

Definitions

  • the present disclosure relates to the technical field of data processing, and in particular, to a battery detection method and device.
  • the first is a purely manual quality inspection, where single crystal silicon solar cells on the production line are manually observed to determine if there are defects.
  • the second is a machine-assisted manual quality inspection method.
  • the machine collects pictures of monocrystalline silicon solar cells on the production line, and the quality inspection system combines defects defined in advance to identify whether there are defects in the pictures.
  • the first method has high labor cost and poor efficiency.
  • the defects defined in the quality inspection system are solidified and difficult to update, and can only identify simple defects and difficult to identify complex defects, which reduces the quality inspection efficiency.
  • the present disclosure aims to solve at least one of the technical problems in the related art.
  • a first object of the present disclosure is to propose a battery detection method for solving the problems of poor battery detection efficiency and high cost in the prior art.
  • a second object of the present disclosure is to propose a battery detection device.
  • a third object of the present disclosure is to propose another battery detection device.
  • a fourth object of the present disclosure is to propose a non-transitory computer-readable storage medium.
  • a fifth object of the present disclosure is to propose a computer program product.
  • an embodiment of the first aspect of the present disclosure provides a battery detection method, including:
  • the detection result includes: whether a defect exists, a type and a position of the defect;
  • a control instruction is sent to a control device of a production node corresponding to the picture, so that the control device shunts a battery corresponding to the defective picture according to the control instruction.
  • the defect detection model is a deep neural network model; the structure of the defect detection model is determined according to a Mask RCNN algorithm.
  • the method before the inputting the picture into a preset defect detection model and obtaining a detection result output by the defect detection model, the method further includes:
  • the training data includes: historical pictures of batteries on the production line and defect annotation results;
  • the defect annotation results include: defect types and defect locations;
  • the trained defect detection model is determined as the preset defect detection model.
  • the method further includes:
  • the method further includes:
  • re-training the defect detection model according to the updated training data includes:
  • the defect detection model is re-trained according to the updated training data.
  • the number of the defect detection models is multiple, and they are respectively set on different servers;
  • the step of inputting the picture into a preset defect detection model and obtaining a detection result output by the defect detection model includes:
  • the picture is input to a first defect detection model, and a detection result output by the first defect detection model is obtained.
  • the battery detection method in the embodiment of the present disclosure obtains pictures of each battery on a battery production line and corresponding production nodes; inputs the pictures into a preset defect detection model, and obtains a detection result output by the defect detection model.
  • the detection result includes: whether or not There is a defect, the type and location of the defect; when the detection result is a defect, a control instruction is sent to the control device of the production node corresponding to the picture, so that the control device shunts the battery corresponding to the defective picture according to the control instruction, thereby passing Combine the defect detection model to identify battery defects, be able to identify simple defects and complex defects, and retrain the defect detection model based on the detection results, so that the defect detection model can identify the most recent defects, and based on the identified defect pairs,
  • the battery performs automatic shunting without manual participation, which improves the efficiency and accuracy of battery detection and greatly reduces labor costs.
  • an embodiment of the second aspect of the present disclosure provides a battery detection device, including:
  • An acquisition module for acquiring pictures of each battery on a battery production line and corresponding production nodes
  • a detection module configured to input the picture into a preset defect detection model and obtain a detection result output by the defect detection model; the detection result includes: whether a defect exists, a type and a position of the defect;
  • a sending module configured to send a control instruction to a control device of a production node corresponding to the picture when the detection result is a defect, so that the control device performs a battery corresponding to the defective picture according to the control instruction Diversion.
  • the defect detection model is a deep neural network model; the structure of the defect detection model is determined according to a Mask RCNN algorithm.
  • the device further includes: a training module and a determination module;
  • the acquisition module is further configured to acquire training data;
  • the training data includes: a historical picture of a battery on a production line and a defect labeling result; and the defect labeling result includes: a defect type and a defect location;
  • the training module is configured to train an initial defect detection model according to the training data until a preset loss function meets a corresponding condition;
  • the determining module is configured to determine the trained defect detection model as the preset defect detection model.
  • the device further includes: an auditing module and an adding module;
  • the auditing module is configured to audit the detection result corresponding to the picture
  • the adding module is configured to add the picture and the corresponding detection result to the training data after the review is passed to obtain updated training data;
  • the training module is further configured to retrain the defect detection model according to the updated training data.
  • the obtaining module is further configured to obtain the number of pictures and corresponding detection results added in the updated training data
  • the training module is specifically configured to re-train the defect detection model according to the updated training data when the number is greater than a preset number threshold.
  • the number of the defect detection models is multiple, and they are respectively set on different servers;
  • the detection module is specifically configured to:
  • the picture is input to a first defect detection model, and a detection result output by the first defect detection model is obtained.
  • the battery detection device in the embodiment of the present disclosure obtains pictures of each battery on a battery production line and corresponding production nodes; inputs the pictures into a preset defect detection model, and obtains detection results output by the defect detection model.
  • the detection results include: whether or not There is a defect, the type and location of the defect; when the detection result is a defect, a control instruction is sent to the control device of the production node corresponding to the picture, so that the control device shunts the battery corresponding to the defective picture according to the control instruction, thereby passing Combine the defect detection model to identify battery defects, be able to identify simple defects and complex defects, and retrain the defect detection model based on the detection results, so that the defect detection model can identify the most recent defects, and based on the identified defect pairs,
  • the battery performs automatic shunting without manual participation, which improves the efficiency and accuracy of battery detection and greatly reduces labor costs.
  • an embodiment of the third aspect of the present disclosure provides another battery detection device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processing When the controller executes the program, the battery detection method as described above is implemented.
  • an embodiment of the fourth aspect of the present disclosure proposes a computer-readable storage medium on which a computer program is stored, which is executed by a processor to implement the battery detection method as described above.
  • an embodiment of the fifth aspect of the present disclosure proposes a computer program product that, when executed by an instruction processor in the computer program product, implements the battery detection method as described above.
  • FIG. 1 is a schematic flowchart of a battery detection method according to an embodiment of the present disclosure
  • FIG. 2 is a schematic flowchart of another battery detection method according to an embodiment of the present disclosure.
  • FIG. 3 is a schematic structural diagram of a battery detection device according to an embodiment of the present disclosure.
  • FIG. 4 is a schematic structural diagram of another battery detection device according to an embodiment of the present disclosure.
  • FIG. 5 is a schematic structural diagram of another battery detection device according to an embodiment of the present disclosure.
  • FIG. 6 is a schematic structural diagram of another battery detection device according to an embodiment of the present disclosure.
  • FIG. 1 is a schematic flowchart of a battery detection method according to an embodiment of the present disclosure. As shown in FIG. 1, the battery detection method includes the following steps:
  • S101 Obtain pictures of each battery on a battery production line, and a corresponding production node.
  • the main body of the battery detection method provided by the present disclosure is a battery detection device.
  • the battery detection device may be a hardware device such as a terminal device, a server, or software installed on the hardware device.
  • the battery involved in this embodiment may be a battery such as a single crystal silicon solar cell.
  • the detection of the battery may be a detection of a specific part of the battery, for example, a detection of a battery electroluminescent (EL).
  • EL battery electroluminescent
  • the picture of the battery may be a picture taken on the battery of each production node on the battery production line.
  • cameras can be set at multiple positions of each production node of the battery production line, for taking pictures of the batteries of each production node and collecting pictures of the batteries.
  • the camera may be a fixed camera or a mobile camera.
  • the picture is input into a preset defect detection model, and a detection result output by the defect detection model is obtained.
  • the detection result includes: whether a defect exists, a type and a position of the defect.
  • the type of the defect may be, for example, a defect such as a crack, a chip, a dummy solder, or a broken grid.
  • the defect detection model can be a deep neural network model; the structure of the defect detection model is determined according to the Mask RCNN algorithm.
  • the structure of the defect detection model may include a convolution layer, a pooling layer, a fully connected layer, and the like.
  • the convolution layer is used to extract features in the picture and generate a feature map corresponding to the picture.
  • the pooling layer is used to perform dimensionality reduction on the feature map, remove non-main features in the feature map, retain the main features in the feature map, and have high robustness to the deformation, blurring, and lighting changes of the picture on the production line.
  • the fully connected layer is a branch of the network based on instance segmentation. This branch is based on feature maps.
  • the feature map is restored to its original size using an algorithm based on bipartite interpolation.
  • Each pixel is predicted to obtain its own instance, such as color and grayscale information, and then each object or part in the picture is known. By comparing each object or component with the object or component under normal conditions, it is possible to determine whether there is a defect in the picture, and the type and location of the defect when there is a defect.
  • the number of defect detection models is multiple, and they are respectively set on different servers.
  • the process of performing step 102 by the battery testing device may specifically be: obtaining the load of each defect detection model; selecting a corresponding first load detection model from each of the defect detection models whose load meets a preset load condition; The first defect detection model acquires a detection result output by the first defect detection model.
  • the setting of multiple defect detection models and the selection of a first defect detection model corresponding to a preset load condition from each of the defect detection models for image detection, compared to a single defect detection model Reduce the workload of the defect detection model, increase the detection speed of pictures, and then improve the detection efficiency of the battery.
  • control device may be, for example, a conveyor belt, a robot arm, or the like, or may be a corresponding controller of the conveyor belt, the robot arm, or the like.
  • a prompt message may be sent to the manager to prompt the manager to manually shunt the battery corresponding to the defective picture.
  • the method may further include: generating a log according to the detection result and the shunt result and storing the log for management personnel to view.
  • the log may include pictures of each battery, picture collection time, detection results, and shunt results.
  • the battery detection method in the embodiment of the present disclosure obtains pictures of each battery on a battery production line and corresponding production nodes; inputs the pictures into a preset defect detection model, and obtains a detection result output by the defect detection model.
  • the detection result includes: whether or not There is a defect, the type and location of the defect; when the detection result is a defect, a control instruction is sent to the control device of the production node corresponding to the picture, so that the control device shunts the battery corresponding to the defective picture according to the control instruction, thereby passing Combine the defect detection model to identify battery defects, be able to identify simple defects and complex defects, and retrain the defect detection model based on the detection results, so that the defect detection model can identify the most recent defects, and based on the identified defect pairs,
  • the battery performs automatic shunting without manual participation, which improves the efficiency and accuracy of battery detection and greatly reduces labor costs.
  • FIG. 2 is a schematic flowchart of a battery detection method according to an embodiment of the present disclosure. As shown in FIG. 2, based on the embodiment shown in FIG. 1, before step 102, the battery detection method may further include the following steps:
  • the training data includes: a historical picture of the battery on the production line and a defect labeling result; and the defect labeling result includes: a defect type and a defect location.
  • the historical picture refers to the historical picture of the battery taken on the production line before the current time.
  • the defect labeling result corresponding to the historical image can be the result obtained by manually labeling the defect type and defect location of the historical image.
  • the preset loss function may be determined according to a loss function of a convolutional layer, a pooling layer, a fully connected layer, and the like in the defect detection model.
  • the condition corresponding to the loss function may be a threshold that the loss function needs to meet.
  • the loss function satisfies the corresponding conditions, which means that the value of the loss function is smaller than the threshold.
  • the method may further include the following steps: reviewing the detection result corresponding to the picture; after the review is passed, adding the picture and the corresponding detection result to the training data To obtain updated training data; retrain the defect detection model based on the updated training data.
  • retraining the defect detection model according to the picture and the corresponding detection result can further improve the accuracy of the defect detection model detection; and when a new defect appears in the detection result, the defect detection model can also detect New defects in pictures.
  • the method before re-training the defect detection model based on the updated training data, the method further includes: obtaining updated training data, the number of pictures added and corresponding detection results; corresponding Re-training the defect detection model according to the updated training data, including: when the number is greater than a preset number threshold, re-training the defect detection model according to the updated training data.
  • the defect detection model in order to reduce the number of times the defect detection model is trained and avoid repeated training, when the number of newly added pictures and corresponding detection results in the training data is greater than a preset number threshold, re-train the defect detection model. .
  • FIG. 3 is a schematic structural diagram of a battery detection device according to an embodiment of the present disclosure. As shown in FIG. 3, it includes: an obtaining module 31, a detecting module 32, and a sending module 33.
  • the obtaining module 31 is configured to obtain pictures of each battery on a battery production line and a corresponding production node;
  • a detection module 32 is configured to input the picture into a preset defect detection model, and obtain a detection result output by the defect detection model.
  • the detection result includes: whether a defect exists, a type and a position of the defect;
  • a sending module 33 is configured to send a control instruction to a control device of a production node corresponding to the picture when the detection result is a defect, so that the control device responds to a battery corresponding to the defective picture according to the control instruction. Perform shunting.
  • the battery detection device provided by the present disclosure may be a hardware device such as a terminal device, a server, or software installed on the hardware device.
  • the battery involved in this embodiment may be a battery such as a single crystal silicon solar cell.
  • the detection of the battery may be a detection of a specific part of the battery, for example, a detection of a battery electroluminescent (EL).
  • EL battery electroluminescent
  • the picture of the battery may be a picture taken on the battery of each production node on the battery production line.
  • cameras can be set at multiple positions of each production node of the battery production line, for taking pictures of the batteries of each production node and collecting pictures of the batteries.
  • the camera may be a fixed camera or a mobile camera.
  • the type of the defect may be, for example, a defect such as a crack, a chip, a dummy solder, or a broken grid.
  • the defect detection model can be a deep neural network model; the structure of the defect detection model is determined according to the Mask RCNN algorithm.
  • the structure of the defect detection model may include a convolution layer, a pooling layer, a fully connected layer, and the like.
  • the convolution layer is used to extract features in the picture and generate a feature map corresponding to the picture.
  • the pooling layer is used to perform dimensionality reduction on the feature map, remove non-main features in the feature map, retain the main features in the feature map, and have high robustness to the deformation, blurring, and lighting changes of the picture on the production line.
  • the fully connected layer is a branch of the network based on instance segmentation. This branch is based on feature maps.
  • the feature map is restored to its original size using an algorithm based on bipartite interpolation.
  • Each pixel is predicted to obtain its own instance, such as color and grayscale information, and then each object or part in the picture is known. By comparing each object or component with the object or component under normal conditions, it is possible to determine whether there is a defect in the picture, and the type and location of the defect when there is a defect.
  • the detection module 32 may be specifically configured to obtain the load amount of each defect detection model; select a first defect detection model with a corresponding load amount satisfying a preset load condition from each defect detection model; input a picture into the first defect detection A model to obtain a detection result output by the first defect detection model.
  • the setting of multiple defect detection models and the selection of a first defect detection model corresponding to a preset load condition from each of the defect detection models for image detection, compared to a single defect detection model Reduce the workload of the defect detection model, increase the detection speed of pictures, and then improve the detection efficiency of the battery.
  • control device may be, for example, a conveyor belt, a robot arm, or the like, or may be a corresponding controller of the conveyor belt, the robot arm, or the like.
  • a prompt message may be sent to the manager to prompt the manager to manually shunt the battery corresponding to the defective picture.
  • the device may further include: a generating module, configured to generate a log according to the detection result and the shunt result, and store the log for management personnel to view.
  • the log may include pictures of each battery, picture collection time, detection results, and shunt results.
  • the battery detection device in the embodiment of the present disclosure obtains pictures of each battery on a battery production line and corresponding production nodes; inputs the pictures into a preset defect detection model, and obtains detection results output by the defect detection model.
  • the detection results include: whether or not There is a defect, the type and location of the defect; when the detection result is a defect, a control instruction is sent to the control device of the production node corresponding to the picture, so that the control device shunts the battery corresponding to the defective picture according to the control instruction, thereby passing Combine the defect detection model to identify battery defects, be able to identify simple defects and complex defects, and retrain the defect detection model based on the detection results, so that the defect detection model can identify the most recent defects, and based on the identified defect pairs,
  • the battery performs automatic shunting without manual participation, which improves the efficiency and accuracy of battery detection and greatly reduces labor costs.
  • the device may further include a training module 34 and a determination module 35.
  • the obtaining module 31 is further configured to obtain training data;
  • the training data includes: a historical picture of a battery on a production line and a defect labeling result; and the defect labeling result includes: a defect type and a defect location;
  • the training module 34 is configured to train an initial defect detection model according to the training data until a preset loss function meets a corresponding condition;
  • the determining module 35 is configured to determine the trained defect detection model as the preset defect detection model.
  • the historical picture refers to the historical picture of the battery taken on the production line before the current time.
  • the defect labeling result corresponding to the historical image can be the result obtained by manually labeling the defect type and defect location of the historical image.
  • the device may further include: an auditing module 36 and an adding module 37.
  • the review module 36 is configured to review the detection result corresponding to the picture
  • the adding module 37 is configured to add the picture and the corresponding detection result to the training data after the review is passed to obtain updated training data;
  • the training module 34 is further configured to retrain the defect detection model according to the updated training data.
  • retraining the defect detection model according to the picture and the corresponding detection result can further improve the accuracy of the defect detection model detection; and when a new defect appears in the detection result, the defect detection model can also detect New defects in pictures.
  • the obtaining module 31 is further configured to obtain the number of pictures and corresponding detection results added to the updated training data; correspondingly, the training module 34 is specific For re-training the defect detection model based on the updated training data when the number is greater than a preset number threshold.
  • the defect detection model in order to reduce the number of times the defect detection model is trained and avoid repeated training, when the number of newly added pictures and corresponding detection results in the training data is greater than a preset number threshold, re-train the defect detection model. .
  • FIG. 6 is a schematic structural diagram of another battery detection device according to an embodiment of the present disclosure.
  • the battery detection device includes:
  • the battery detection device further includes:
  • the communication interface 1003 is used for communication between the memory 1001 and the processor 1002.
  • the memory 1001 is configured to store a computer program that can be run on the processor 1002.
  • the memory 1001 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), for example, at least one disk memory.
  • the processor 1002 is configured to implement the battery detection method according to the foregoing embodiment when the program is executed.
  • the communication interface 1003, the memory 1001, and the processor 1002 may be connected to each other through a bus and complete communication with each other.
  • the bus may be an Industry Standard Architecture (ISA) bus, an External Device Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Wait.
  • ISA Industry Standard Architecture
  • PCI External Device Component
  • EISA Extended Industry Standard Architecture
  • the bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only a thick line is used in FIG. 6, but it does not mean that there is only one bus or one type of bus.
  • the memory 1001, the processor 1002, and the communication interface 1003 are integrated and implemented on a chip, the memory 1001, the processor 1002, and the communication interface 1003 can complete communication with each other through an internal interface.
  • the processor 1002 may be a central processing unit (CPU), or a specific integrated circuit (ASIC), or may be configured to implement one or more embodiments of the present disclosure. integrated circuit.
  • CPU central processing unit
  • ASIC application specific integrated circuit
  • the present disclosure also provides a non-transitory computer-readable storage medium on which a computer program is stored, which when executed by a processor implements the battery detection method as described above.
  • the present disclosure also provides a computer program product that, when executed by an instruction processor in the computer program product, implements the battery detection method as described above.
  • first and second are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as “first” and “second” may explicitly or implicitly include at least one of the features. In the description of the present disclosure, the meaning of "a plurality” is at least two, for example, two, three, etc., unless it is specifically and specifically defined otherwise.
  • any process or method description in a flowchart or otherwise described herein can be understood as representing a module, fragment, or portion of code that includes one or more executable instructions for implementing steps of a custom logic function or process
  • the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order according to the functions involved, which should It is understood by those skilled in the art to which the embodiments of the present disclosure belong.
  • Logic and / or steps represented in a flowchart or otherwise described herein, for example, a sequenced list of executable instructions that may be considered to implement a logical function, may be embodied in any computer-readable medium, For use by, or in combination with, an instruction execution system, device, or device (such as a computer-based system, a system that includes a processor, or another system that can fetch and execute instructions from an instruction execution system, device, or device) Or equipment.
  • a "computer-readable medium” may be any device that can contain, store, communicate, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.
  • computer-readable media include the following: electrical connections (electronic devices) with one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), Read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disk read-only memory (CDROM).
  • the computer-readable medium may even be paper or other suitable medium on which the program can be printed, because, for example, by optically scanning the paper or other medium, followed by editing, interpretation, or other suitable Processing to obtain the program electronically and then store it in computer memory.
  • portions of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof.
  • multiple steps or methods may be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system.
  • Discrete logic circuits with logic gates for implementing logic functions on data signals Logic circuits, ASICs with suitable combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.
  • a person of ordinary skill in the art can understand that all or part of the steps carried by the methods in the foregoing embodiments can be implemented by a program instructing related hardware.
  • the program can be stored in a computer-readable storage medium.
  • the program is When executed, one or a combination of the steps of the method embodiment is included.
  • each functional unit in each embodiment of the present disclosure may be integrated into one processing module, or each unit may exist separately physically, or two or more units may be integrated into one module.
  • the above integrated modules may be implemented in the form of hardware or software functional modules. When the integrated module is implemented in the form of a software function module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.
  • the aforementioned storage medium may be a read-only memory, a magnetic disk, or an optical disk.

Landscapes

  • Engineering & Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Immunology (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Manufacturing & Machinery (AREA)
  • Electrochemistry (AREA)
  • General Chemical & Material Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
  • Secondary Cells (AREA)
  • Testing Or Measuring Of Semiconductors Or The Like (AREA)

Abstract

一种电池检测方法及装置，该方法包括：获取电池生产线上各个电池的图片，以及对应的生产节点；将图片输入预设的缺陷检测模型，获取缺陷检测模型输出的检测结果；检测结果中包括：是否存在缺陷、缺陷的类型以及位置；在检测结果为存在缺陷时，向图片对应的生产节点的控制设备发送控制指令，以使控制设备根据控制指令对存在缺陷的图片对应的电池进行分流。

Description

电池检测方法及装置
相关申请的交叉引用
本公开基于申请号为201810980598.3，申请日为2018年8月27日的中国专利申请提出，并要求该中国专利申请的优先权，该中国专利申请的全部内容在此引入本申请作为参考。
技术领域
本公开涉及数据处理技术领域，尤其涉及一种电池检测方法及装置。
背景技术
目前对单晶硅太阳能电池的质量检测方法有两种。第一种是纯人工质检，由人工对生产线上的单晶硅太阳能电池进行观察，确定是否有缺陷。第二种是机器辅助的人工质检方式，由机器来采集生产线上的单晶硅太阳能电池的图片，由质检系统结合提前定义好的缺陷来识别图片中是否有缺陷。
第一种方法，人力成本高，效率差。第二种方法中，质检系统中定义的缺陷是固化的，难以进行更新，且只能识别简单的缺陷，难以识别复杂的缺陷，降低了质量检测效率。
发明内容
本公开旨在至少在一定程度上解决相关技术中的技术问题之一。
为此，本公开的第一个目的在于提出一种电池检测方法，用于解决现有技术中电池检测效率差，成本高的问题。
本公开的第二个目的在于提出一种电池检测装置。
本公开的第三个目的在于提出另一种电池检测装置。
本公开的第四个目的在于提出一种非临时性计算机可读存储介质。
本公开的第五个目的在于提出一种计算机程序产品。
为达上述目的，本公开第一方面实施例提出了一种电池检测方法，包括：
获取电池生产线上各个电池的图片，以及对应的生产节点；
将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果；所述检测结果中包括：是否存在缺陷、缺陷的类型以及位置；
在所述检测结果为存在缺陷时，向所述图片对应的生产节点的控制设备发送控制指令， 以使所述控制设备根据所述控制指令对存在缺陷的图片对应的电池进行分流。
进一步的，所述缺陷检测模型为深度神经网络模型；所述缺陷检测模型的结构根据Mask RCNN算法确定。
进一步的，所述将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果之前，还包括：
获取训练数据；所述训练数据中包括：生产线上电池的历史图片以及缺陷标注结果；所述缺陷标注结果中包括：缺陷类型以及缺陷位置；
根据所述训练数据，对初始的缺陷检测模型进行训练，直至预设的损失函数满足对应的条件；
将训练好的缺陷检测模型，确定为所述预设的缺陷检测模型。
进一步的，所述将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果之后，还包括：
对所述图片对应的检测结果进行审核；
审核通过后，将所述图片以及对应的检测结果添加到所述训练数据中，得到更新后的训练数据；
根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
进一步的，根据更新后的训练数据，对所述缺陷检测模型重新进行训练之前，还包括：
获取所述更新后的训练数据中，添加的图片以及对应的检测结果的数量；
对应的，根据更新后的训练数据，对所述缺陷检测模型重新进行训练，包括：
在所述数量大于预设数量阈值时，根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
进一步的，所述缺陷检测模型的数量为多个，分别设置在不同的服务器上；
所述将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果，包括：
获取各个缺陷检测模型的负载量；
从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型；
将所述图片输入第一缺陷检测模型，获取所述第一缺陷检测模型输出的检测结果。
本公开实施例的电池检测方法，通过获取电池生产线上各个电池的图片，以及对应的生产节点；将图片输入预设的缺陷检测模型，获取缺陷检测模型输出的检测结果；检测结果中包括：是否存在缺陷、缺陷的类型以及位置；在检测结果为存在缺陷时，向图片对应的生产节点的控制设备发送控制指令，以使控制设备根据控制指令对存在缺陷的图片对应的电池进 行分流，从而通过结合缺陷检测模型来识别电池的缺陷，能够识别简单缺陷和复杂缺陷，且能够结合检测结果对缺陷检测模型进行再次训练，使得缺陷检测模型能够识别到最新出现的缺陷，以及根据识别到的缺陷对电池进行自动分流，不需要人工参与，提高了电池检测的效率和准确度，大大降低了人工成本。
为达上述目的，本公开第二方面实施例提出了一种电池检测装置，包括：
获取模块，用于获取电池生产线上各个电池的图片，以及对应的生产节点；
检测模块，用于将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果；所述检测结果中包括：是否存在缺陷、缺陷的类型以及位置；
发送模块，用于在所述检测结果为存在缺陷时，向所述图片对应的生产节点的控制设备发送控制指令，以使所述控制设备根据所述控制指令对存在缺陷的图片对应的电池进行分流。
进一步的，所述缺陷检测模型为深度神经网络模型；所述缺陷检测模型的结构根据Mask RCNN算法确定。
进一步的，所述的装置还包括：训练模块和确定模块；
所述获取模块，还用于获取训练数据；所述训练数据中包括：生产线上电池的历史图片以及缺陷标注结果；所述缺陷标注结果中包括：缺陷类型以及缺陷位置；
所述训练模块，用于根据所述训练数据，对初始的缺陷检测模型进行训练，直至预设的损失函数满足对应的条件；
所述确定模块，用于将训练好的缺陷检测模型，确定为所述预设的缺陷检测模型。
进一步的，所述的装置还包括：审核模块和添加模块；
所述审核模块，用于对所述图片对应的检测结果进行审核；
所述添加模块，用于在审核通过后，将所述图片以及对应的检测结果添加到所述训练数据中，得到更新后的训练数据；
所述训练模块，还用于根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
进一步的，所述获取模块，还用于获取所述更新后的训练数据中，添加的图片以及对应的检测结果的数量；
对应的，所述训练模块具体用于，在所述数量大于预设数量阈值时，根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
进一步的，所述缺陷检测模型的数量为多个，分别设置在不同的服务器上；
对应的，所述检测模块具体用于，
获取各个缺陷检测模型的负载量；
从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型；
将所述图片输入第一缺陷检测模型，获取所述第一缺陷检测模型输出的检测结果。
本公开实施例的电池检测装置，通过获取电池生产线上各个电池的图片，以及对应的生产节点；将图片输入预设的缺陷检测模型，获取缺陷检测模型输出的检测结果；检测结果中包括：是否存在缺陷、缺陷的类型以及位置；在检测结果为存在缺陷时，向图片对应的生产节点的控制设备发送控制指令，以使控制设备根据控制指令对存在缺陷的图片对应的电池进行分流，从而通过结合缺陷检测模型来识别电池的缺陷，能够识别简单缺陷和复杂缺陷，且能够结合检测结果对缺陷检测模型进行再次训练，使得缺陷检测模型能够识别到最新出现的缺陷，以及根据识别到的缺陷对电池进行自动分流，不需要人工参与，提高了电池检测的效率和准确度，大大降低了人工成本。
为达上述目的，本公开第三方面实施例提出了另一种电池检测装置，包括：存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现如上所述的电池检测方法。
为了实现上述目的，本公开第四方面实施例提出了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现如上所述的电池检测方法。
为了实现上述目的，本公开第五方面实施例提出了一种计算机程序产品，当所述计算机程序产品中的指令处理器执行时，实现如上所述的电池检测方法。
本公开附加的方面和优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本公开的实践了解到。
附图说明
本公开上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：
图1为本公开实施例提供的一种电池检测方法的流程示意图；
图2为本公开实施例提供的另一种电池检测方法的流程示意图；
图3为本公开实施例提供的一种电池检测装置的结构示意图；
图4为本公开实施例提供的另一种电池检测装置的结构示意图；
图5为本公开实施例提供的另一种电池检测装置的结构示意图；
图6为本公开实施例提供的另一种电池检测装置的结构示意图。
具体实施方式
下面详细描述本公开的实施例，所述实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，旨在用于解释本公开，而不能理解为对本公开的限制。
下面参考附图描述本公开实施例的电池检测方法及装置。
图1为本公开实施例提供的一种电池检测方法的流程示意图。如图1所示，该电池检测方法包括以下步骤：
S101、获取电池生产线上各个电池的图片，以及对应的生产节点。
本公开提供的电池检测方法的执行主体为电池检测装置，电池检测装置可以为终端设备、服务器等硬件设备，或者为硬件设备上安装的软件。本实施例中涉及到的电池可以为单晶硅太阳能电池等电池。对电池的检测可以为针对电池特定部件的检测，例如针对电池电致发光部件(Electroluminescent，EL)的检测。
本实施例中，电池的图片，可以为在电池生产线上，对各个生产节点的电池拍摄得到的图片。其中，可以在电池生产线的各个生产节点的多个位置上设置摄像头，用于对各个生产节点的电池进行拍摄，采集电池的图片。其中，摄像头可以为固定摄像头，也可以为移动摄像头。
S102、将图片输入预设的缺陷检测模型，获取缺陷检测模型输出的检测结果；检测结果中包括：是否存在缺陷、缺陷的类型以及位置。
本实施例中，缺陷的类型例如可以为，隐裂、碎片、虚焊、断栅等缺陷。缺陷检测模型具体可以为深度神经网络模型；缺陷检测模型的结构根据Mask RCNN算法确定。本实施例中，缺陷检测模型的结构可以包括：卷积层、池化层、全连接层等。卷积层用于提取图片中的特征，生成与图片对应的特征图。池化层用于对特征图进行降维操作，去除特征图中的非主要特征，保留特征图中的主要特征，对生产线上图片的变形、模糊、光照变化等具有较高的鲁棒性。全连接层为基于实例分割的网络分支。该分支基于特征图，利用基于二分插值的算法将特征图还原到原图大小，对每一个像素进行预测得到其所属的实例，例如颜色、灰度等信息，进而获知图片中的各物体或者部件，将各物体或者部件与正常情况下的物体或者部件进行比对，从而能够确定图片中是否存在缺陷，在存在缺陷时缺陷的类型以及位置。
进一步的，在上述实施例的基础上，缺陷检测模型的数量为多个，分别设置在不同的服务器上。对应的，电池检测装置执行步骤102的过程具体可以为，获取各个缺陷检测模型的负载量；从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型；将图片输入第一缺陷检测模型，获取所述第一缺陷检测模型输出的检测结果。
本实施例中，多个缺陷检测模型的设置，以及从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型进行图像检测，相对于单个缺陷检测模型来说，能够降低缺陷检测模型的工作量，提高图片的检测速度，进而提高电池的检测效率。
S103、在检测结果为存在缺陷时，向图片对应的生产节点的控制设备发送控制指令，以使控制设备根据控制指令对存在缺陷的图片对应的电池进行分流。
本实施例中，控制设备例如可以为传送带、机械臂等，或者可以为传送带、机械臂等对应的控制器。另外，为了确保对存在缺陷的图片对应的电池的分流效果，在检测结果为存在缺陷时，可以向管理人员发送提示信息，提示管理人员对存在缺陷的图片对应的电池进行手动分流。
另外，还需要进行说明的是，步骤103之后，所述的方法还可以包括：根据检测结果以及分流结果，生成日志并进行存储，以便管理人员进行查看。其中，日志中可以包括：各个电池的图片、图片采集时间、检测结果以及分流结果等。
本公开实施例的电池检测方法，通过获取电池生产线上各个电池的图片，以及对应的生产节点；将图片输入预设的缺陷检测模型，获取缺陷检测模型输出的检测结果；检测结果中包括：是否存在缺陷、缺陷的类型以及位置；在检测结果为存在缺陷时，向图片对应的生产节点的控制设备发送控制指令，以使控制设备根据控制指令对存在缺陷的图片对应的电池进行分流，从而通过结合缺陷检测模型来识别电池的缺陷，能够识别简单缺陷和复杂缺陷，且能够结合检测结果对缺陷检测模型进行再次训练，使得缺陷检测模型能够识别到最新出现的缺陷，以及根据识别到的缺陷对电池进行自动分流，不需要人工参与，提高了电池检测的效率和准确度，大大降低了人工成本。
图2为本公开实施例提供的一种电池检测方法的流程示意图。如图2所示，在图1所示实施例的基础上，步骤102之前，所述的电池检测方法还可以包括以下步骤：
S104、获取训练数据；训练数据中包括：生产线上电池的历史图片以及缺陷标注结果；缺陷标注结果中包括：缺陷类型以及缺陷位置。
本实施例中，历史图片指的是，当前时刻之前在生产线上拍摄得到的电池的历史图片。历史图片对应的缺陷标注结果，可以为人工对历史图片进行缺陷类型以及缺陷位置进行标注后得到的结果。
S105、根据训练数据，对初始的缺陷检测模型进行训练，直至预设的损失函数满足对应的条件。
本实施例中，预设的损失函数可以根据缺陷检测模型中卷积层、池化层、全连接层等的 损失函数确定。其中，损失函数对应的条件可以为，损失函数所需要满足的阈值。损失函数满足对应的条件，指的是损失函数的值小于所述阈值。
S106、将训练好的缺陷检测模型，确定为预设的缺陷检测模型。
进一步的，在上述实施例的基础上，步骤102之后，所述的方法还可以包括以下步骤：对图片对应的检测结果进行审核；审核通过后，将图片以及对应的检测结果添加到训练数据中，得到更新后的训练数据；根据更新后的训练数据，对缺陷检测模型重新进行训练。
本实施例中，根据图片以及对应的检测结果，对缺陷检测模型重新进行训练，能够进一步提高缺陷检测模型检测的准确度；且当检测结果中出现新的缺陷时，缺陷检测模型也能够检测出图片中新的缺陷。
进一步的，在上述实施例的基础上，根据更新后的训练数据，对缺陷检测模型重新进行训练之前，还包括：获取更新后的训练数据中，添加的图片以及对应的检测结果的数量；对应的，根据更新后的训练数据，对缺陷检测模型重新进行训练，包括：在数量大于预设数量阈值时，根据更新后的训练数据，对缺陷检测模型重新进行训练。
本实施例中，为了减少缺陷检测模型的训练次数，避免重复训练，可以在训练数据中新添加的图片以及对应的检测结果的数量大于预设数量阈值时，再启动对缺陷检测模型的重新训练。
图3为本公开实施例提供的一种电池检测装置的结构示意图。如图3所示，包括：获取模块31、检测模块32和发送模块33。
其中，获取模块31，用于获取电池生产线上各个电池的图片，以及对应的生产节点；
检测模块32，用于将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果；所述检测结果中包括：是否存在缺陷、缺陷的类型以及位置；
发送模块33，用于在所述检测结果为存在缺陷时，向所述图片对应的生产节点的控制设备发送控制指令，以使所述控制设备根据所述控制指令对存在缺陷的图片对应的电池进行分流。
本公开提供的电池检测装置可以为终端设备、服务器等硬件设备，或者为硬件设备上安装的软件。本实施例中涉及到的电池可以为单晶硅太阳能电池等电池。对电池的检测可以为针对电池特定部件的检测，例如针对电池电致发光部件(Electroluminescent，EL)的检测。
本实施例中，电池的图片，可以为在电池生产线上，对各个生产节点的电池拍摄得到的图片。其中，可以在电池生产线的各个生产节点的多个位置上设置摄像头，用于对各个生产节点的电池进行拍摄，采集电池的图片。其中，摄像头可以为固定摄像头，也可以为移动摄 像头。
本实施例中，缺陷的类型例如可以为，隐裂、碎片、虚焊、断栅等缺陷。缺陷检测模型具体可以为深度神经网络模型；缺陷检测模型的结构根据Mask RCNN算法确定。本实施例中，缺陷检测模型的结构可以包括：卷积层、池化层、全连接层等。卷积层用于提取图片中的特征，生成与图片对应的特征图。池化层用于对特征图进行降维操作，去除特征图中的非主要特征，保留特征图中的主要特征，对生产线上图片的变形、模糊、光照变化等具有较高的鲁棒性。全连接层为基于实例分割的网络分支。该分支基于特征图，利用基于二分插值的算法将特征图还原到原图大小，对每一个像素进行预测得到其所属的实例，例如颜色、灰度等信息，进而获知图片中的各物体或者部件，将各物体或者部件与正常情况下的物体或者部件进行比对，从而能够确定图片中是否存在缺陷，在存在缺陷时缺陷的类型以及位置。
进一步的，在上述实施例的基础上，缺陷检测模型的数量为多个，分别设置在不同的服务器上。对应的，检测模块32具体可以用于，获取各个缺陷检测模型的负载量；从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型；将图片输入第一缺陷检测模型，获取所述第一缺陷检测模型输出的检测结果。
本实施例中，多个缺陷检测模型的设置，以及从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型进行图像检测，相对于单个缺陷检测模型来说，能够降低缺陷检测模型的工作量，提高图片的检测速度，进而提高电池的检测效率。
本实施例中，控制设备例如可以为传送带、机械臂等，或者可以为传送带、机械臂等对应的控制器。另外，为了确保对存在缺陷的图片对应的电池的分流效果，在检测结果为存在缺陷时，可以向管理人员发送提示信息，提示管理人员对存在缺陷的图片对应的电池进行手动分流。
另外，还需要进行说明的是，所述的装置还可以包括：生成模块，用于根据检测结果以及分流结果，生成日志并进行存储，以便管理人员进行查看。其中，日志中可以包括：各个电池的图片、图片采集时间、检测结果以及分流结果等。
本公开实施例的电池检测装置，通过获取电池生产线上各个电池的图片，以及对应的生产节点；将图片输入预设的缺陷检测模型，获取缺陷检测模型输出的检测结果；检测结果中包括：是否存在缺陷、缺陷的类型以及位置；在检测结果为存在缺陷时，向图片对应的生产节点的控制设备发送控制指令，以使控制设备根据控制指令对存在缺陷的图片对应的电池进行分流，从而通过结合缺陷检测模型来识别电池的缺陷，能够识别简单缺陷和复杂缺陷，且能够结合检测结果对缺陷检测模型进行再次训练，使得缺陷检测模型能够识别到最新出现的缺陷，以及根据识别到的缺陷对电池进行自动分流，不需要人工参与，提高了电池检测的效 率和准确度，大大降低了人工成本。
进一步的，结合参考图4，在图3所示实施例的基础上，所述的装置还可以包括：训练模块34和确定模块35。
其中，所述获取模块31，还用于获取训练数据；所述训练数据中包括：生产线上电池的历史图片以及缺陷标注结果；所述缺陷标注结果中包括：缺陷类型以及缺陷位置；
所述训练模块34，用于根据所述训练数据，对初始的缺陷检测模型进行训练，直至预设的损失函数满足对应的条件；
所述确定模块35，用于将训练好的缺陷检测模型，确定为所述预设的缺陷检测模型。
本实施例中，历史图片指的是，当前时刻之前在生产线上拍摄得到的电池的历史图片。历史图片对应的缺陷标注结果，可以为人工对历史图片进行缺陷类型以及缺陷位置进行标注后得到的结果。
进一步的，结合参考图5，在图4所示实施例的基础上，所述的装置还可以包括：审核模块36和添加模块37。
其中，所述审核模块36，用于对所述图片对应的检测结果进行审核；
所述添加模块37，用于在审核通过后，将所述图片以及对应的检测结果添加到所述训练数据中，得到更新后的训练数据；
所述训练模块34，还用于根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
本实施例中，根据图片以及对应的检测结果，对缺陷检测模型重新进行训练，能够进一步提高缺陷检测模型检测的准确度；且当检测结果中出现新的缺陷时，缺陷检测模型也能够检测出图片中新的缺陷。
进一步的，在上述实施例的基础上，所述获取模块31，还用于获取所述更新后的训练数据中，添加的图片以及对应的检测结果的数量；对应的，所述训练模块34具体用于，在所述数量大于预设数量阈值时，根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
本实施例中，为了减少缺陷检测模型的训练次数，避免重复训练，可以在训练数据中新添加的图片以及对应的检测结果的数量大于预设数量阈值时，再启动对缺陷检测模型的重新训练。
图6为本公开实施例提供的另一种电池检测装置的结构示意图。该电池检测装置包括：
存储器1001、处理器1002及存储在存储器1001上并可在处理器1002上运行的计算机程序。
处理器1002执行所述程序时实现上述实施例中提供的电池检测方法。
进一步地，电池检测装置还包括：
通信接口1003，用于存储器1001和处理器1002之间的通信。
存储器1001，用于存放可在处理器1002上运行的计算机程序。
存储器1001可能包含高速RAM存储器，也可能还包括非易失性存储器(non-volatile memory)，例如至少一个磁盘存储器。
处理器1002，用于执行所述程序时实现上述实施例所述的电池检测方法。
如果存储器1001、处理器1002和通信接口1003独立实现，则通信接口1003、存储器1001和处理器1002可以通过总线相互连接并完成相互间的通信。所述总线可以是工业标准体系结构(Industry Standard Architecture，简称为ISA)总线、外部设备互连(Peripheral Component，简称为PCI)总线或扩展工业标准体系结构(Extended Industry Standard Architecture，简称为EISA)总线等。所述总线可以分为地址总线、数据总线、控制总线等。为便于表示，图6中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
可选的，在具体实现上，如果存储器1001、处理器1002及通信接口1003，集成在一块芯片上实现，则存储器1001、处理器1002及通信接口1003可以通过内部接口完成相互间的通信。
处理器1002可能是一个中央处理器(Central Processing Unit，简称为CPU)，或者是特定集成电路(Application Specific Integrated Circuit，简称为ASIC)，或者是被配置成实施本公开实施例的一个或多个集成电路。
本公开还提供一种非临时性计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现如上所述的电池检测方法。
本公开还提供一种计算机程序产品，当所述计算机程序产品中的指令处理器执行时，实现如上所述的电池检测方法。
在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本公开的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不必须针对的是相同的实施例或示例。而且，描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外，在不相互矛盾的情况下，本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者 隐含地包括至少一个该特征。在本公开的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。
流程图中或在此以其他方式描述的任何过程或方法描述可以被理解为，表示包括一个或更多个用于实现定制逻辑功能或过程的步骤的可执行指令的代码的模块、片段或部分，并且本公开的优选实施方式的范围包括另外的实现，其中可以不按所示出或讨论的顺序，包括根据所涉及的功能按基本同时的方式或按相反的顺序，来执行功能，这应被本公开的实施例所属技术领域的技术人员所理解。
在流程图中表示或在此以其他方式描述的逻辑和/或步骤，例如，可以被认为是用于实现逻辑功能的可执行指令的定序列表，可以具体实现在任何计算机可读介质中，以供指令执行系统、装置或设备(如基于计算机的系统、包括处理器的系统或其他可以从指令执行系统、装置或设备取指令并执行指令的系统)使用，或结合这些指令执行系统、装置或设备而使用。就本说明书而言，"计算机可读介质"可以是任何可以包含、存储、通信、传播或传输程序以供指令执行系统、装置或设备或结合这些指令执行系统、装置或设备而使用的装置。计算机可读介质的更具体的示例(非穷尽性列表)包括以下：具有一个或多个布线的电连接部(电子装置)，便携式计算机盘盒(磁装置)，随机存取存储器(RAM)，只读存储器(ROM)，可擦除可编辑只读存储器(EPROM或闪速存储器)，光纤装置，以及便携式光盘只读存储器(CDROM)。另外，计算机可读介质甚至可以是可在其上打印所述程序的纸或其他合适的介质，因为可以例如通过对纸或其他介质进行光学扫描，接着进行编辑、解译或必要时以其他合适方式进行处理来以电子方式获得所述程序，然后将其存储在计算机存储器中。
应当理解，本公开的各部分可以用硬件、软件、固件或它们的组合来实现。在上述实施方式中，多个步骤或方法可以用存储在存储器中且由合适的指令执行系统执行的软件或固件来实现。如，如果用硬件来实现和在另一实施方式中一样，可用本领域公知的下列技术中的任一项或他们的组合来实现：具有用于对数据信号实现逻辑功能的逻辑门电路的离散逻辑电路，具有合适的组合逻辑门电路的专用集成电路，可编程门阵列(PGA)，现场可编程门阵列(FPGA)等。
本技术领域的普通技术人员可以理解实现上述实施例方法携带的全部或部分步骤是可以通过程序来指令相关的硬件完成，所述的程序可以存储于一种计算机可读存储介质中，该程序在执行时，包括方法实施例的步骤之一或其组合。
此外，在本公开各个实施例中的各功能单元可以集成在一个处理模块中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。所述集成的模块如果以软件功 能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。
上述提到的存储介质可以是只读存储器，磁盘或光盘等。
以上所述仅是本公开的优选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本公开原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本公开的保护范围。

Claims (15)

  1. 一种电池检测方法，其特征在于，包括：
    获取电池生产线上各个电池的图片，以及对应的生产节点；
    将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果；所述检测结果中包括：是否存在缺陷、缺陷的类型以及位置；
    在所述检测结果为存在缺陷时，向所述图片对应的生产节点的控制设备发送控制指令，以使所述控制设备根据所述控制指令对存在缺陷的图片对应的电池进行分流。
  2. 根据权利要求1所述的方法，其特征在于，所述缺陷检测模型为深度神经网络模型；所述缺陷检测模型的结构根据Mask RCNN算法确定。
  3. 根据权利要求1或2所述的方法，其特征在于，所述将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果之前，还包括：
    获取训练数据；所述训练数据中包括：生产线上电池的历史图片以及缺陷标注结果；所述缺陷标注结果中包括：缺陷类型以及缺陷位置；
    根据所述训练数据，对初始的缺陷检测模型进行训练，直至预设的损失函数满足对应的条件；
    将训练好的缺陷检测模型，确定为所述预设的缺陷检测模型。
  4. 根据权利要求3所述的方法，其特征在于，所述将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果之后，还包括：
    对所述图片对应的检测结果进行审核；
    审核通过后，将所述图片以及对应的检测结果添加到所述训练数据中，得到更新后的训练数据；
    根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
  5. 根据权利要求4所述的方法，其特征在于，根据更新后的训练数据，对所述缺陷检测模型重新进行训练之前，还包括：
    获取所述更新后的训练数据中，添加的图片以及对应的检测结果的数量；
    对应的，根据更新后的训练数据，对所述缺陷检测模型重新进行训练，包括：
    在所述数量大于预设数量阈值时，根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
  6. 根据权利要求1-5任一项所述的方法，其特征在于，所述缺陷检测模型的数量为多 个，分别设置在不同的服务器上；
    所述将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果，包括：
    获取各个缺陷检测模型的负载量；
    从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型；
    将所述图片输入第一缺陷检测模型，获取所述第一缺陷检测模型输出的检测结果。
  7. 一种电池检测装置，其特征在于，包括：
    获取模块，用于获取电池生产线上各个电池的图片，以及对应的生产节点；
    检测模块，用于将所述图片输入预设的缺陷检测模型，获取所述缺陷检测模型输出的检测结果；所述检测结果中包括：是否存在缺陷、缺陷的类型以及位置；
    发送模块，用于在所述检测结果为存在缺陷时，向所述图片对应的生产节点的控制设备发送控制指令，以使所述控制设备根据所述控制指令对存在缺陷的图片对应的电池进行分流。
  8. 根据权利要求7所述的装置，其特征在于，所述缺陷检测模型为深度神经网络模型；所述缺陷检测模型的结构根据Mask RCNN算法确定。
  9. 根据权利要求7或8所述的装置，其特征在于，还包括：训练模块和确定模块；
    所述获取模块，还用于获取训练数据；所述训练数据中包括：生产线上电池的历史图片以及缺陷标注结果；所述缺陷标注结果中包括：缺陷类型以及缺陷位置；
    所述训练模块，用于根据所述训练数据，对初始的缺陷检测模型进行训练，直至预设的损失函数满足对应的条件；
    所述确定模块，用于将训练好的缺陷检测模型，确定为所述预设的缺陷检测模型。
  10. 根据权利要求9所述的装置，其特征在于，还包括：审核模块和添加模块；
    所述审核模块，用于对所述图片对应的检测结果进行审核；
    所述添加模块，用于在审核通过后，将所述图片以及对应的检测结果添加到所述训练数据中，得到更新后的训练数据；
    所述训练模块，还用于根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
  11. 根据权利要求10所述的装置，其特征在于，
    所述获取模块，还用于获取所述更新后的训练数据中，添加的图片以及对应的检测结果的数量；
    对应的，所述训练模块具体用于，在所述数量大于预设数量阈值时，根据更新后的训练数据，对所述缺陷检测模型重新进行训练。
  12. 根据权利要求7-11任一项所述的装置，其特征在于，所述缺陷检测模型的数量为 多个，分别设置在不同的服务器上；
    对应的，所述检测模块具体用于，
    获取各个缺陷检测模型的负载量；
    从各个缺陷检测模型中选择对应的负载量满足预设负载条件的第一缺陷检测模型；
    将所述图片输入第一缺陷检测模型，获取所述第一缺陷检测模型输出的检测结果。
  13. 一种电池检测装置，其特征在于，包括：
    存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现如权利要求1-6中任一所述的电池检测方法。
  14. 一种非临时性计算机可读存储介质，其上存储有计算机程序，其特征在于，该程序被处理器执行时实现如权利要求1-6中任一所述的电池检测方法。
  15. 一种计算机程序产品，当所述计算机程序产品中的指令处理器执行时，实现如权利要求1-6中任一所述的电池检测方法。
PCT/CN2019/093383 2018-08-27 2019-06-27 电池检测方法及装置 WO2020042741A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP19853678.1A EP3683572B1 (en) 2018-08-27 2019-06-27 Battery detection method and device
US16/650,279 US11158044B2 (en) 2018-08-27 2019-06-27 Battery detection method and device
KR1020207036130A KR102494946B1 (ko) 2018-08-27 2019-06-27 배터리 검출 방법 및 장치(battery detection method and device)
JP2020519278A JP7018135B2 (ja) 2018-08-27 2019-06-27 バッテリー検出方法及び装置

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810980598.3 2018-08-27
CN201810980598.3A CN109239075B (zh) 2018-08-27 2018-08-27 电池检测方法及装置

Publications (1)

Publication Number Publication Date
WO2020042741A1 true WO2020042741A1 (zh) 2020-03-05

Family

ID=65068351

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/093383 WO2020042741A1 (zh) 2018-08-27 2019-06-27 电池检测方法及装置

Country Status (6)

Country Link
US (1) US11158044B2 (zh)
EP (1) EP3683572B1 (zh)
JP (1) JP7018135B2 (zh)
KR (1) KR102494946B1 (zh)
CN (1) CN109239075B (zh)
WO (1) WO2020042741A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022102075A1 (ja) * 2020-11-13 2022-05-19 日本電気株式会社 学習装置、処理装置、学習方法、処理方法及びプログラム
CN114638294A (zh) * 2022-03-10 2022-06-17 深圳市腾盛精密装备股份有限公司 一种数据增强方法、装置、终端设备及存储介质
CN117115162A (zh) * 2023-10-24 2023-11-24 中安芯界控股集团有限公司 基于视觉分析对电池进行检测的芯片生产控制系统

Families Citing this family (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109239075B (zh) * 2018-08-27 2021-11-30 北京百度网讯科技有限公司 电池检测方法及装置
CN111563869B (zh) * 2019-01-25 2023-07-21 宁波舜宇光电信息有限公司 用于摄像模组质检的污点测试方法
CN112465814A (zh) * 2020-12-17 2021-03-09 无锡日联科技股份有限公司 基于深度学习的电池overhang的计算方法和装置
CN114913108A (zh) * 2021-01-29 2022-08-16 京东方科技集团股份有限公司 显示基板的图像分类方法和装置
CN112967245B (zh) * 2021-03-01 2022-11-08 创新奇智(青岛)科技有限公司 电池检测方法、装置、电子设备及可读存储介质
US20220318667A1 (en) * 2021-03-30 2022-10-06 Accenture Global Solutions Limited Intelligent real-time defect prediction, detection, and ai driven automated correction solution
CN113487567B (zh) * 2021-07-05 2023-07-07 广东奥普特科技股份有限公司 尺寸精度的检测方法、装置、电子设备以及存储介质
CN113567446B (zh) * 2021-07-06 2022-07-19 北京东方国信科技股份有限公司 组件缺陷检测质量分级的方法及系统
CN114152422B (zh) * 2021-12-02 2024-04-05 国家石油天然气管网集团有限公司 一种在油气管道开挖现场管道非外部缺陷类型的检测方法
CN116547661A (zh) * 2021-12-03 2023-08-04 宁德时代新能源科技股份有限公司 用于图像分割的数据标注方法和系统以及图像分割装置
CN114418963B (zh) * 2021-12-28 2023-12-05 淮阴工学院 一种基于机器视觉的电池极板缺陷检测方法
WO2023167498A1 (ko) * 2022-03-02 2023-09-07 주식회사 엘지에너지솔루션 배터리 불량 검출 장치, 방법 및 시스템
CN115829906B (zh) * 2022-04-18 2024-01-30 宁德时代新能源科技股份有限公司 对复合料带的阴极极片进行缺陷检测的方法、装置、系统
CN117501507A (zh) * 2022-04-25 2024-02-02 宁德时代新能源科技股份有限公司 电池极片的Overhang检测方法、装置、设备、存储介质
CN115049592A (zh) * 2022-05-19 2022-09-13 广州超音速自动化科技股份有限公司 基于三维斜拍的电池焊盘检测方法、系统、设备及介质
CN115829913B (zh) * 2022-08-10 2024-05-17 宁德时代新能源科技股份有限公司 裸电芯外观检测方法、装置、计算机设备和存储介质
CN115825070A (zh) * 2022-08-10 2023-03-21 宁德时代新能源科技股份有限公司 电芯检测方法、装置、计算机设备和介质
WO2024049201A1 (ko) * 2022-08-31 2024-03-07 주식회사 엘지에너지솔루션 배터리 전극 검사 장치 및 방법
CN115829921B (zh) * 2022-09-16 2024-01-05 宁德时代新能源科技股份有限公司 检测电芯缺陷的方法、装置和计算机可读存储介质
CN118130464A (zh) * 2022-12-02 2024-06-04 华为技术有限公司 一种缺陷检测方法、装置及系统
CN115937184B (zh) * 2022-12-28 2024-07-05 中轻长泰(长沙)智能科技股份有限公司 一种物品装配检测方法、装置及终端设备
CN116542980B (zh) * 2023-07-06 2023-11-03 宁德时代新能源科技股份有限公司 缺陷检测方法、装置、计算机设备、存储介质和程序产品
CN118469918A (zh) * 2024-01-17 2024-08-09 杭州广立微电子股份有限公司 晶圆缺陷检测的可视化优化方法、装置和计算机设备
CN117635605B (zh) * 2024-01-23 2024-06-18 宁德时代新能源科技股份有限公司 电池目检确认方法、装置、电子设备和存储介质

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6266138B1 (en) * 1999-10-12 2001-07-24 Perceptron, Inc. System and method for detecting defects in a surface of a workpiece
CN204936176U (zh) * 2015-09-10 2016-01-06 天津滨海祥宝塑料制品有限责任公司 一种塑料桶生产线的自动检漏系统
CN206107748U (zh) * 2016-09-13 2017-04-19 国药集团冯了性（佛山）药业有限公司 药酒自动化筛查分流包装生产线
CN108090897A (zh) * 2017-12-18 2018-05-29 川亿电脑(深圳)有限公司 印刷线路板缺陷的检测方法、检测装置及存储介质
CN108230317A (zh) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 钢板缺陷检测分类方法、装置、设备及计算机可读介质
CN108320278A (zh) * 2018-01-09 2018-07-24 北京百度网讯科技有限公司 产品缺陷检测定位方法、装置、设备及计算机可读介质
CN109239075A (zh) * 2018-08-27 2019-01-18 北京百度网讯科技有限公司 电池检测方法及装置

Family Cites Families (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4377786A (en) * 1980-08-06 1983-03-22 Radco Industries, Inc. Battery testing apparatus
JP2006243923A (ja) * 2005-03-01 2006-09-14 Meiwa E Tec:Kk ワーク追跡管理装置
CN101546557B (zh) * 2008-03-28 2011-03-23 展讯通信（上海）有限公司 用于音频内容识别的分类器参数更新方法
CN101539533B (zh) * 2009-03-11 2011-06-22 华南理工大学 电池内部缺陷自动检测装置及方法
CN103502899B (zh) * 2011-01-26 2016-09-28 谷歌公司 动态预测建模平台
US8595154B2 (en) * 2011-01-26 2013-11-26 Google Inc. Dynamic predictive modeling platform
CN103020642B (zh) * 2012-10-08 2016-07-13 江苏省环境监测中心 水环境监测质控数据分析方法
JP6061713B2 (ja) * 2013-02-08 2017-01-18 本田技研工業株式会社 検査装置、検査方法及びプログラム
CN203705352U (zh) * 2014-01-23 2014-07-09 四川大学 一种基于机器视觉的磁瓦在线检测设备
CN204241383U (zh) * 2014-08-19 2015-04-01 广州伊索自动化科技有限公司 一种基于机器视觉的汽车连接件检测系统
CN105118044B (zh) * 2015-06-16 2017-11-07 华南理工大学 一种轮形铸造产品缺陷自动检测方法
CN105205479A (zh) * 2015-10-28 2015-12-30 小米科技有限责任公司 人脸颜值评估方法、装置及终端设备
CN105279382B (zh) * 2015-11-10 2017-12-22 成都数联易康科技有限公司 一种医疗保险异常数据在线智能检测方法
CN105352967A (zh) * 2015-11-17 2016-02-24 浙江集英工业智能机器技术有限公司 一种软磁磁芯自动检测用的多工位传输摄像机构
US9965901B2 (en) * 2015-11-19 2018-05-08 KLA—Tencor Corp. Generating simulated images from design information
JP6661398B2 (ja) 2016-02-03 2020-03-11 キヤノン株式会社 情報処理装置および情報処理方法
JPWO2017179712A1 (ja) * 2016-04-14 2019-03-14 大日本印刷株式会社 電池用包装材料、その製造方法、電池用包装材料の成形時における不良判定方法、アルミニウム合金箔
CN109791111B (zh) * 2016-08-22 2022-01-11 丘比株式会社 食品检查装置、食品检查方法以及食品检查装置的识别机构的学习方法
US10115040B2 (en) * 2016-09-14 2018-10-30 Kla-Tencor Corporation Convolutional neural network-based mode selection and defect classification for image fusion
CN106530284A (zh) * 2016-10-21 2017-03-22 广州视源电子科技股份有限公司 基于图像识别的焊点类型检测和装置
CN106568783B (zh) * 2016-11-08 2019-12-03 广东工业大学 一种五金零件缺陷检测系统及方法
CN106814088A (zh) * 2016-12-30 2017-06-09 镇江苏仪德科技有限公司 基于机器视觉对电池片颜色分选的检测装置及方法
CN106875381B (zh) * 2017-01-17 2020-04-28 同济大学 一种基于深度学习的手机外壳缺陷检测方法
CN107192759B (zh) * 2017-06-09 2019-08-27 湖南大学 一种基于感应光热辐射的光伏电池无损检测方法及系统
CN107831173A (zh) * 2017-10-17 2018-03-23 哈尔滨工业大学（威海） 光伏组件缺陷检测方法及系统
CN107833220B (zh) * 2017-11-28 2021-06-11 河海大学常州校区 基于深度卷积神经网络与视觉显著性的织物缺陷检测方法
CN108061735A (zh) * 2017-12-01 2018-05-22 工业互联网创新中心（上海）有限公司 零部件表面缺陷的识别方法和装置
CN108305243B (zh) * 2017-12-08 2021-11-30 五邑大学 一种基于深度学习的磁瓦表面缺陷检测方法
CN108074231B (zh) * 2017-12-18 2020-04-21 浙江工业大学 一种基于卷积神经网络的磁片表面缺陷检测方法
CN108257121B (zh) * 2018-01-09 2019-01-25 北京百度网讯科技有限公司 产品缺陷检测模型更新的方法、装置、存储介质及终端设备
CN108154508B (zh) * 2018-01-09 2019-05-24 北京百度网讯科技有限公司 产品缺陷检测定位的方法、装置、存储介质及终端设备

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6266138B1 (en) * 1999-10-12 2001-07-24 Perceptron, Inc. System and method for detecting defects in a surface of a workpiece
CN204936176U (zh) * 2015-09-10 2016-01-06 天津滨海祥宝塑料制品有限责任公司 一种塑料桶生产线的自动检漏系统
CN206107748U (zh) * 2016-09-13 2017-04-19 国药集团冯了性（佛山）药业有限公司 药酒自动化筛查分流包装生产线
CN108090897A (zh) * 2017-12-18 2018-05-29 川亿电脑(深圳)有限公司 印刷线路板缺陷的检测方法、检测装置及存储介质
CN108230317A (zh) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 钢板缺陷检测分类方法、装置、设备及计算机可读介质
CN108320278A (zh) * 2018-01-09 2018-07-24 北京百度网讯科技有限公司 产品缺陷检测定位方法、装置、设备及计算机可读介质
CN109239075A (zh) * 2018-08-27 2019-01-18 北京百度网讯科技有限公司 电池检测方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3683572A4

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022102075A1 (ja) * 2020-11-13 2022-05-19 日本電気株式会社 学習装置、処理装置、学習方法、処理方法及びプログラム
JP7439953B2 (ja) 2020-11-13 2024-02-28 日本電気株式会社 学習装置、処理装置、学習方法、処理方法及びプログラム
CN114638294A (zh) * 2022-03-10 2022-06-17 深圳市腾盛精密装备股份有限公司 一种数据增强方法、装置、终端设备及存储介质
CN117115162A (zh) * 2023-10-24 2023-11-24 中安芯界控股集团有限公司 基于视觉分析对电池进行检测的芯片生产控制系统

Also Published As

Publication number Publication date
EP3683572A1 (en) 2020-07-22
CN109239075A (zh) 2019-01-18
EP3683572B1 (en) 2022-06-01
US20210209739A1 (en) 2021-07-08
JP7018135B2 (ja) 2022-02-09
EP3683572A4 (en) 2021-06-02
KR102494946B1 (ko) 2023-02-06
CN109239075B (zh) 2021-11-30
JP2020536328A (ja) 2020-12-10
US11158044B2 (en) 2021-10-26
KR20210010541A (ko) 2021-01-27

Similar Documents

Publication Publication Date Title
WO2020042741A1 (zh) 电池检测方法及装置
WO2020239015A1 (zh) 一种图像识别、分类方法、装置、电子设备及存储介质
US11010285B2 (en) Fault detection and localization to generate failing test cases using combinatorial test design techniques
CN111507958A (zh) 目标检测方法、检测模型的训练方法及电子设备
CN107749091A (zh) 配电巡检方法及终端设备
TWI729606B (zh) 用於一邊緣運算網路的負載平衡裝置及方法
TW201947463A (zh) 模型測試的方法及裝置
JP6800351B2 (ja) 電極シートのバリを検出するための方法および装置
CN111310746B (zh) 文本行检测方法、模型训练方法、装置、服务器及介质
WO2020010638A1 (zh) 一种图像坏点检测方法及装置
CN115937170A (zh) 电路板检测方法、装置、计算机设备及存储介质
CN110969600A (zh) 一种产品缺陷检测方法、装置、电子设备及存储介质
CN111931799B (zh) 图像识别方法及装置
CN113077416A (zh) 一种基于图像处理的焊点焊接缺陷检测方法与系统
JP7059889B2 (ja) 学習装置、画像生成装置、学習方法、及び学習プログラム
CN109087280A (zh) 连铸坯质量检测方法、装置、控制设备、服务器及介质
CN116895009A (zh) 模型训练方法、油雾去除方法、装置、设备及存储介质
CN115908308A (zh) 基于偏振光成像的表面缺陷检测方法、设备及存储介质
CN108357193A (zh) 丝印检测方法、装置、终端设备及存储介质
WO2024026990A1 (zh) 识别模型的自动迭代训练方法、系统、设备和存储介质
CN117129029B (zh) 一种芯片检测方法及系统
CN113255911B (zh) 一种模型训练方法、装置、电子设备及存储介质
CN114820543B (zh) 缺陷检测方法和装置
CN116258902A (zh) 绝缘子巡检图像的识别方法、装置、设备及存储介质
KR20240013924A (ko) 전력설비 결함 진단 시스템 및 방법

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2020519278

Country of ref document: JP

Kind code of ref document: A

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19853678

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2019853678

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2019853678

Country of ref document: EP

Effective date: 20200416

ENP Entry into the national phase

Ref document number: 20207036130

Country of ref document: KR

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE