Target detection method and device, electronic equipment, storage medium and program product

Abstract

The invention provides a target detection method and device, electronic equipment, a storage medium and a program product, and the method comprises the steps: carrying out the tracking shooting of a to-be-detected object in response to the detection of the to-be-detected object, and obtaining a detection video; performing first target analysis according to the detection video; performing frame picture extraction on the detection video according to a set rule to obtain a detection picture; performing second target analysis according to the detection picture; and in response to existence of the first target and / or the second target, determining that the to-be-detected object is a target object.

Description

Technical Field

[0001] This disclosure relates to the field of hardware testing technology, and in particular to a target testing method, apparatus, electronic device, storage medium, and program product. Background Technology

[0002] In existing processes, before panels for structures such as displays leave the factory, they generally need to undergo quality inspection. Related technologies involve taking static photos of the panels and examining these images to determine if any defects are present.

[0003] However, this detection method can only detect static panels and cannot detect panels in other scenarios, making the detection limited and incomplete, and the reliability of the detection needs to be improved.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] In view of this, this disclosure proposes a target detection method, apparatus, electronic device, storage medium, and program product to solve or partially solve the above-mentioned problems.

[0006] In view of the above objectives, firstly, this disclosure provides a target detection method, comprising:

[0007] In response to the detection of an object to be inspected, the object to be inspected is tracked and photographed to obtain a detection video;

[0008] Perform first target analysis based on the detected video;

[0009] Frame images are extracted from the detected video according to the set rules to obtain the detected images;

[0010] Perform a second target analysis based on the detected image;

[0011] In response to the presence of the first target and / or the second target, the object to be inspected is determined to be a target object.

[0012] In some exemplary embodiments, after performing the first target analysis based on the detected video, the method further includes:

[0013] In response to the presence of the first target, the time period in which the first target appears in the detected video is determined, and the image is cropped according to the time period to obtain at least one target image.

[0014] In some exemplary embodiments, the step of extracting frame images from the detected video according to a set rule to obtain a detected image includes:

[0015] The duration of the detection video is determined, and the number and location of the detection images are determined based on the duration. The detection images are then extracted based on the number and location.

[0016] In some exemplary embodiments, after determining that the object to be inspected is the target object, the method further includes:

[0017] Determine the target image corresponding to the target object;

[0018] The target image is segmented to obtain at least two fragment images;

[0019] The at least two fragment images are stored in at least two nodes of the network; wherein the network comprises a plurality of nodes for data storage.

[0020] In some exemplary embodiments, the segmentation process of the target image includes:

[0021] Determine the target location in the target image, and crop the image based on the target location;

[0022] The segmentation process is performed based on the truncation results.

[0023] In some exemplary embodiments, storing the at least two fragment images in at least two nodes of the network includes:

[0024] Generate a corresponding image key for any fragment image, and encrypt the fragment image according to the image key to obtain an encrypted image;

[0025] Determine the public and private keys of the node storing any of the encrypted images, and encrypt the image key using the public key to obtain the encryption key;

[0026] Perform a hash calculation on any of the encrypted images, and sign the hash calculation result according to the private key to obtain a signature result;

[0027] The encrypted image, the encryption key, and the signature result are stored as relevant data for any fragment image.

[0028] In some exemplary embodiments, the network includes a private blockchain network.

[0029] In some exemplary embodiments, after storing the at least two fragment images in at least two nodes of the network, the method further includes:

[0030] In response to the request to obtain the target image, the at least two nodes determine the permission to obtain the at least two fragment images.

[0031] In response to the requirement that the number of fragment images can be obtained in greater than a set threshold according to the obtained permission, the target image is formed based on the at least two fragment images, thereby responding to the obtained request.

[0032] Based on the same concept, in a second aspect, this disclosure also provides a target detection device, comprising:

[0033] The first module is used to track and capture images of the object under inspection in response to the detection of the object under inspection, thereby obtaining a detection video.

[0034] The second module is used to perform a first target analysis based on the detected video;

[0035] The third module is used to extract frame images from the detected video according to the set rules to obtain the detected images;

[0036] The fourth module is used to perform a second target analysis based on the detected image;

[0037] The fifth module is used to determine the object to be inspected as a target object in response to the presence of the first target and / or the second target.

[0038] Based on the same concept, in a third aspect, this disclosure also provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any of the preceding claims.

[0039] Based on the same concept, in a fourth aspect, this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method as described in any of the preceding claims.

[0040] Based on the same concept, in a fifth aspect, this disclosure also provides a computer program product including computer program instructions that, when run on a computer, cause the computer to perform the method as described in any of the preceding claims.

[0041] As described above, this disclosure provides a target detection method, apparatus, electronic device, storage medium, and program product. After identifying the object to be inspected, this disclosure first captures a video of the object to be inspected, obtaining a dynamic inspection video. Then, it extracts frames from the inspection video to form inspection images. This allows for dynamic inspection of the inspection video and static inspection of the inspection images, employing a dual-channel inspection mode of video and images. The images, with their high resolution, clearly present the subtle static structure of the object to be inspected; the video, on the other hand, records the dynamic process of the object to be inspected in real time, accurately capturing dynamic issues such as intermittent faults and displacement of moving parts. The two complement each other, comprehensively covering various targets, significantly improving detection accuracy, ensuring the integrity of the detection results, and providing a more reliable basis for product quality control. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments or related technologies of this disclosure, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram illustrating an exemplary application scenario provided by an embodiment of this disclosure.

[0044] Figure 2 A flowchart illustrating an exemplary method provided in an embodiment of this disclosure.

[0045] Figure 3 This is a schematic diagram illustrating the process of adding a new node to an exemplary blockchain provided in this disclosure embodiment.

[0046] Figure 4 A schematic diagram of the structure of an exemplary device provided in an embodiment of this disclosure.

[0047] Figure 5 This is a schematic diagram of the electronic device structure provided in an embodiment of this disclosure. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this specification clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0049] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this disclosure should have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms "first," "second," and similar terms used in the embodiments of this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element, object, or method step preceding the term covers the element, object, or method step listed after the term and its equivalents, but does not exclude other elements, objects, or method steps. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0050] As described in the background section, related technologies utilize Automated Optical Inspection (AOI) equipment for panel quality inspection. However, relying solely on image detection makes it difficult to identify and predict dynamic defects, leading to frequent missed detections. Furthermore, panel or glass cannot be captured and identified during assembly line transport, rendering this time unusable and impacting production efficiency. Moreover, relying solely on image detection has other drawbacks. For example, when storing a single image, the data is centralized, and if the storage medium fails (e.g., hard drive failure), the entire image data is easily lost, severely affecting the traceability and analysis of inspection results. Additionally, storing a large number of images consumes significant storage space, resulting in high storage costs and low efficiency when retrieving specific images. On the other hand, from an inspection perspective, static images struggle to capture dynamic changes. For defects that change over time, such as intermittent circuit failures or dynamic component displacement, traditional image detection methods face a serious risk of missed detection.

[0051] In light of the above-mentioned practical situation, this disclosure provides a target detection method. After identifying the object to be inspected, the method first captures a video of the object to be inspected, obtaining a dynamic detection video. Then, it extracts frames from the detection video to form detection images. This allows for dynamic detection of the detection video and static detection of the detection images, employing a dual-channel detection mode of video and images. The images, with their high resolution, clearly present the subtle static structure of the object to be inspected; the video, on the other hand, records the dynamic process of the object to be inspected in real time, accurately capturing dynamic issues such as intermittent faults and displacement of moving parts. The two complement each other, comprehensively covering various targets, significantly improving detection accuracy, ensuring the integrity of the detection results, and providing a more reliable basis for product quality control.

[0052] Furthermore, Figure 1 The illustration shows an exemplary application scenario of a target detection method provided by an embodiment of this disclosure.

[0053] refer to Figure 1 In this application scenario, a local terminal device 101 and a server 102 may be included. The local terminal device 101 and the server 102 can be connected via a wired or wireless communication network to achieve data interaction.

[0054] Local terminal device 101 can be a terminal device located close to the user side, possessing data transmission and multimedia input / output functions, such as a desktop computer, mobile computer, or tablet computer. In this application scenario, for example... Figure 1 As shown, the local terminal device 101 can connect to some external devices, such as a camera 103 for image acquisition and a webcam 104 for image processing. Of course, in other application scenarios, it can also be other terminal devices capable of performing the above functions.

[0055] Server 102 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides 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, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0056] In some exemplary embodiments, the target detection method may run on a local terminal device 101 or a server 102.

[0057] When the target detection method runs on server 102, server 102 provides services to users of terminal devices. Local terminal device 101 has a client installed that communicates with server 102. Local terminal device 101 can use the corresponding client to send data to server 102 to determine whether target detection is needed, data transmission, how to acquire data, or how to perform target detection. During the service provision process, server 102, in response to detecting a target object, tracks and captures the target object to obtain a detection video; performs a first target analysis based on the detection video; extracts frame images from the detection video according to set rules to obtain detection images; performs a second target analysis based on the detection images; and, in response to the existence of the first target and / or the second target, determines the target object as a target object.

[0058] In practical implementation, when the target detection method runs on server 102, the method can be implemented and executed based on the cloud interaction system.

[0059] In the above embodiments, the target detection method is described using server 102 as an example. However, this disclosure is not limited thereto. In some exemplary embodiments, the target detection method can also be run on local terminal device 101.

[0060] The local terminal device 101 may include a processor. A client or detection-related program may be installed on the local terminal device 101. The local terminal device 101 can use this client or program to monitor data generated by the camera 103 and webcam 104, and detect objects to be inspected according to target detection methods. In other embodiments, the objects to be inspected may be stored in files such as video files. The local terminal device 101 can use a network video recorder 105 (NVR) or similar device to read the video files, determine the objects to be inspected, and then directly use the corresponding video as the detection video. During the execution of the corresponding process, in response to the detection of an object to be inspected, the local terminal device 101 tracks and captures the object to be inspected to obtain a detection video; performs a first target analysis based on the detection video; extracts frame images from the detection video according to set rules to obtain detection images; performs a second target analysis based on the detection images; and, in response to the existence of the first target and / or the second target, determines the object to be inspected as a target object.

[0061] Finally, there can be multiple local terminal devices 101, and multiple local terminal devices 101 can be connected through wired or wireless communication networks to form a network, and the local terminal device 101 can be a node in the network.

[0062] The following is combined with Figure 1 The application scenarios described above are used to illustrate the target detection method according to exemplary embodiments of this disclosure. It should be noted that the above application scenarios are shown only to facilitate understanding of the spirit and principles of this disclosure, and the embodiments of this disclosure are not limited in any way. Rather, the embodiments of this disclosure can be applied to any applicable scenario.

[0063] Figure 2 An exemplary flowchart of a target detection method provided in an embodiment of this disclosure is shown.

[0064] refer to Figure 2 The present disclosure provides a target detection method, which specifically includes the following steps:

[0065] Step 202: In response to the detection of the object to be inspected, the object to be inspected is tracked and photographed to obtain a detection video.

[0066] In this step, the object to be inspected is the target object that needs to be inspected, such as a panel or glass plate in a specific scenario. In some embodiments, the object to be inspected can be transported via an assembly line or conveyor belt, keeping it in a continuous dynamic state. When it enters the range of the local terminal device 101, the local terminal device 101 can use devices such as a camera 103 or a webcam 104 to track and capture images of the object to be inspected, thereby forming an inspection video. In other embodiments, the local terminal device 101 can also directly acquire videos containing the object to be inspected, and by analyzing the video, it can be determined that it contains the object to be inspected, and then the video can be considered as the inspection video.

[0067] In specific application scenarios, the parameters of the devices connected to the local terminal device 101 can be initialized in advance, and the camera 104 can be set to operate at a high frame rate (e.g., greater than or equal to 10fps). Then, in scenarios where an inspection object is transported using a production line, attributes such as the production line length and the speed of the object being inspected can be determined in advance. This allows the determination of the time it takes for the object to pass through the production line, ultimately determining the length of the inspection video. In other words, each recording will acquire an inspection video of a relatively fixed length.

[0068] In some more specific application scenarios, when a panel (the object to be inspected) enters the pipeline, the relevant controller device sends an entry signal and the panel number (panelID, etc.) to the local terminal device 101. The local terminal device 101 can then recognize the panel entry signal and start pulling the video stream. If no panel enters, it will not pull the stream, thus reducing the system load. When the panel leaves the pipeline, the relevant controller device sends a departure signal to the local terminal device 101. This component can then recognize the panel departure signal and stop pulling the video stream.

[0069] Step 204: Perform first target analysis based on the detected video.

[0070] In this step, the primary objective can be determined based on the specific scenario. For example, the primary objective could be a defect, but the types of defects may differ for different products or objects. Therefore, the analysis of the primary objective can be determined based on the object to be inspected. After determining the primary objective, a corresponding training set can be determined for each primary objective. The training set is then used to train the model, ultimately enabling the relevant model to identify and analyze the primary objective in the object to be inspected.

[0071] Then, taking the first target as an example of a defect, it can also be divided into dynamic defects, static defects, etc.

[0072] In specific application scenarios, static defects can include: Dark Spots (MPP04): Localized areas on or inside the glass panel where abnormal light reflection, transmission, or scattering occurs due to minute impurities, uneven structural density, surface scratches, wear, or contamination, resulting in a significantly darker visual effect compared to the surrounding normal area. Low Bright Spots (MPP08): Areas on the glass panel with brightness lower than normal areas, but not completely dark like dark spots; they are relatively dark but still have some light reflection or transmission. This may be due to slight inhomogeneities within or on the surface of the glass, causing slight changes in light propagation or reflection, resulting in a slightly lower visual brightness. Bright Spots (MPP03): Localized areas on the glass panel that are significantly brighter than the surrounding area. This is usually due to protrusions or foreign objects on the glass surface, or the presence of substances with a different refractive index than the glass matrix, causing strong reflection or refraction of light in these areas, thus forming bright spots. White Gap (MPG02): Generally refers to white, crack-like defects appearing on the glass panel. This may be due to loose bonding between adjacent parts during the glass manufacturing process, resulting in tiny gaps. Light is scattered or reflected at these gaps, creating the appearance of white gaps. It could also be due to missing or uneven sealing material at the glass edges, causing light to leak through the gaps and creating the visual effect of white gaps. White Spot MPF01: White dot-like defects appearing on the glass panel. This may be due to tiny white impurity particles, such as dust or oxides, on or inside the glass surface. These particles strongly scatter light, creating visually white dot-like areas. Additionally, tiny air bubbles inside the glass may also appear as white spots at certain angles because the refractive index of the bubbles differs from that of the glass matrix, causing light scattering and reflection. Thin Line MLL06: Thin, elongated, linear defects appearing on the glass panel surface, with a relatively narrow width. This may be due to mechanical scratches, mold marks, or uneven stress distribution within the glass during the manufacturing process, resulting in these thin, linear defects on the glass surface. The presence of thin lines can affect the strength and appearance quality of the glass, especially in applications with high aesthetic requirements, where thin lines are one of the defects that need strict control. Foreign matter (MBD19): Any foreign material not belonging to the glass matrix present on the glass panel, such as metal shavings, plastic particles, fibers, etc. These foreign matter may have been introduced during the glass production process or adhered to the glass surface during subsequent processing, transportation, and storage. The presence of foreign matter not only affects the appearance of the glass but may also adversely affect its performance, such as reducing its transparency and affecting its optical uniformity. Crack (MAP21): Cracks appearing inside or on the surface of the glass, possibly caused by stress concentration, mechanical impact, or thermal stress. Bubble (MOB01): Gas voids present inside the glass, usually formed when gas is not completely expelled during the glass melting process.Scratch MBA07: Linear damage marks on the glass surface caused by external force.

[0073] Dynamic defects can include: **Moving Cracks (MBD16):** Cracks that appear on or inside the glass during the glass panel production process. Their length, width, or shape may expand or change with time, external force, temperature, and other factors. For example, during glass handling, small cracks may gradually lengthen and widen due to vibration or impact. **Liquid Residual Flow (MAZ03):** Liquids on the glass panel surface, such as cleaning fluids and coating fluids, exhibit irregular flow patterns under the influence of gravity, surface tension, or other external forces. Their shape and position change continuously over time. **Foreign Object Movement (MPF08):** Foreign objects present on or inside the glass panel, such as dust, debris, and metal particles, may change position or move under the influence of vibration, airflow, or other external forces during the production process. **Dynamic Changes in Thermal Deformation (MBA06):** After undergoing high-temperature processing, glass panels may experience thermal deformation due to uneven internal stress distribution or inconsistent cooling processes. This thermal deformation may continue to change over a certain period of time with temperature changes or the release of internal stress. Intermittent Adsorbent MBD04: In the production environment of glass panels, some tiny objects may intermittently adsorb onto the glass surface. These adsorbents may appear and disappear on the glass surface due to static electricity, magnetic fields or other physicochemical effects, or their adsorption positions may change abruptly.

[0074] In practical applications, after obtaining the inspection video, it can be analyzed locally or uploaded to a relevant analysis platform or server for analysis. The relevant model of the local terminal device 101 or server 102 can identify the first target based on training. The first target can be the aforementioned dynamic defect, or it can include some obvious static defects, such as obvious scratches or damage. After detecting the first target, it can be directly marked, relevant information can be recorded, and relevant alarm information can be generated. As for the object to be inspected, it can be used as the target object (such as a defective panel), thus skipping other inspection processes and directly entering the subsequent stages.

[0075] In some embodiments, since the data stream of a video is generally larger than that of an image, after identifying the first target through video detection, the video can be further processed to facilitate storage and transmission. Only a portion of the frames can be extracted for recording the first target. For example, images can be cropped based on the duration of the first target's appearance in the video, and these cropped images can be marked and stored as target images corresponding to the first target for later problem screening and determination of repair and adjustment plans. Simultaneously, uploading only the target images can significantly reduce data transmission volume, effectively alleviate network pressure, ensure fast and stable data transmission under limited bandwidth, and enable timely feedback of detection results to the production process, meeting the production line's stringent real-time requirements.

[0076] In practical applications, after identifying the first target, the time point of its first appearance can be recorded. The first target is then continuously monitored (e.g., its movement trajectory and morphological changes) until it disappears or the video ends, and the time point of disappearance or end is recorded. Throughout this process, the time position of the defect in the video stream can be marked in real time, forming a dynamic defect lifecycle timeline. Subsequently, a target image extraction strategy can be based on the start frame, intermediate frames, and end frame of the first target's appearance. The start frame is the video frame corresponding to the time point of the first target's first appearance, recording the initial state and appearance position of the first target. The intermediate frame can be extracted by calculating the midpoint of the duration; this frame is used to present typical forms in the development process of the first target, such as the intermediate stage of crack propagation or the key trajectory of liquid flow. The end frame can be the last frame before the first target is displayed, recording the final state of the first target, which can help analyze the cause of its disappearance or subsequent impact. Of course, in other embodiments, the selection of target images can be determined according to the specific scenario, without specific limitations. That is, in some embodiments, after performing the first target analysis based on the detected video, the method further includes: in response to the existence of the first target, determining the time period in which the first target appears in the detected video, and performing image cropping based on the time period to obtain at least one target image.

[0077] Step 206: Extract frame images from the detection video according to the set rules to obtain detection images.

[0078] In this step, images can be extracted from the detection video according to certain rules. For example, the extraction method can be determined based on the length of the entire video. The extracted images are the detection images. It should be noted that step 306 can be executed according to the normal process; it can also be performed simultaneously with step 304; or, after the analysis is completed in step 304, the decision to proceed with step 306 can be made based on the analysis results. If the first target is identified in step 304, then the object to be inspected can be directly identified as the target object, thus skipping step 306, etc.

[0079] In some embodiments, for the frame image extraction method of the detection video, the video time can be determined first, for example, by calculating using the aforementioned pipeline length and running speed. Then, frame images can be extracted according to time. (1) If the video time is short, for example, the time is ≤2 seconds, then only 1 frame image can be extracted during the operation of the object to be inspected. The timing starts when the object to be inspected just enters the pipeline. When the object to be inspected runs to the middle position (after time t / 2), the image at that moment is collected as the detection image. (2) If the video time is sufficient, for example, the time is >2 seconds, then an image can be extracted at certain intervals, for example, an image can be extracted at 2-second intervals. The timing starts when the object to be inspected enters the pipeline. When the timing reaches 2 seconds, 1 frame is extracted. Then, 1 frame is extracted every 2 seconds until the object to be inspected leaves the detection area. Specifically, the image of the object to be inspected is collected at each time the timing reaches a multiple of 2 seconds. Then, in order to further improve the representativeness of the detection image, the frames in the middle position of the detection video can be extracted to form the detection image. That is, the object to be inspected is located at the middle position of the pipeline. When the object's running distance reaches half the pipeline length, an additional frame is extracted. This can be achieved by calculating the running distance in real time during the object's movement, and capturing the image at the moment the running distance equals half the pipeline length. However, this method overlaps significantly with images captured in shorter videos, so it is more suitable for scenarios with sufficient video length. Specifically, in some embodiments, the step of extracting frame images from the detection video according to set rules to obtain detection images includes: determining the duration of the detection video, determining the number and location of the detection images to be extracted based on the duration, and extracting the detection images based on the number and location. The number and location of the images can both be determined using the aforementioned method, leveraging the video duration.

[0080] Step 208: Perform a second target analysis based on the detected image.

[0081] In this step, the second target is similar to the first target mentioned above. However, since the second target targets the inspection image, it can correspond to a static target, such as a static defect. That is, the first target analysis mainly identifies dynamic defects, while the second target analysis mainly identifies static defects. Similarly, after detecting the second target, it can be directly marked, relevant information recorded, and related alarm information generated. The object to be inspected can then be used as the target object (e.g., a defective panel), allowing the process to skip other inspection steps and proceed directly to subsequent stages.

[0082] Step 210: In response to the presence of the first target and / or the second target, determine the object to be inspected as the target object.

[0083] In this step, after the first target and / or the second target are identified through analysis, the object to be inspected can be identified as the target object, and subsequent operations can be performed accordingly. For example, the detection results can be output, displayed, or played on a corresponding device to provide feedback to the operator. Of course, in some other embodiments, the output method of the detection results is not limited to display; it can also be used to store, display, use, or reprocess the detection results. The specific output method of the detection results can be flexibly selected according to different application scenarios and implementation needs.

[0084] Specifically, for example, in the application scenario where the method of this embodiment is executed on a single device, the detection results can be directly output on the display component (monitor, projector, etc.) of the current device, so that the operator of the current device can directly see the content of the detection results on the display component.

[0085] For example, in application scenarios where the method of this embodiment is executed on a system composed of multiple devices, the detection results can be sent to other preset devices within the system, i.e., synchronization terminals, as receivers, via any data communication method (wired connection, NFC, Bluetooth, Wi-Fi, cellular network, etc.), so that the synchronization terminals can perform subsequent processing. Optionally, the synchronization terminal can be a preset server, which is generally located in the cloud and serves as a data processing and storage center, capable of storing and distributing the detection results; wherein, the receivers of the distribution are the terminal devices, and the owners or operators of these terminal devices can be relevant personnel involved in the production, supervision, testing, design, etc., of the object to be inspected.

[0086] For example, in the application scenario where the method of this embodiment is executed on a system composed of multiple devices, the detection results can be directly sent to a preset terminal device through any data communication method. The terminal device can be one or more of the devices listed in the preceding paragraphs.

[0087] In one application scenario, once a target object is identified, the identified first and / or second targets need to be stored and recorded to facilitate adjustments to the target object and optimization of the production process. For example, in panel production, whenever a defective target object is detected, information such as the type, location, and size of the defect is recorded, and data acquisition stores relevant video frames or images. Subsequently, based on the severity of the defect, corresponding processing instructions can be issued, such as marking and sorting severely defective target objects off the production line, and recording normal objects to be inspected and allowing them to proceed to the next process.

[0088] In some embodiments, in conjunction with the foregoing, for target objects, the first target and / or the second target are generally recorded by recording target images. This can be done by extracting a portion of frame images from the detection video as target images; or by directly using the detection image as the target image, etc. Regarding the storage of target images, in practical applications, if a single image is used for storage, the data is centralized. If the storage medium fails, such as a hard drive failure, the entire image data can easily be lost, severely affecting the traceability and analysis of detection results. Therefore, a distributed approach can be used for target image storage. For example, the image can be segmented according to certain rules (e.g., uniform segmentation or segmentation according to certain weights) to obtain multiple fragment images. Then, using a network formed by multiple local terminal devices 101 (nodes), the multiple fragment images are stored in multiple nodes in the network, for example, each node stores only one or a few fragment images. Therefore, through reasonable segmentation and storage strategies, not only is the overall storage space occupied effectively reduced, but advanced indexing and verification mechanisms are also used to ensure data security and reliability, facilitating fast retrieval and access. Simultaneously, by uploading only a portion of the video stream's frame images, the amount of data stored can be significantly reduced, alleviating the burden on the storage system, significantly optimizing the data management process, and improving data utilization efficiency. Specifically, in some embodiments, after determining the object to be inspected as the target object, the method further includes: determining the target image corresponding to the target object; segmenting the target image to obtain at least two fragment images; and storing the at least two fragment images in at least two nodes of a network; wherein the network includes multiple nodes for data storage.

[0089] Furthermore, since the target image may contain too much irrelevant content—for example, when photographing an object to be inspected on an assembly line, the entire object may be captured first, followed by images of its surroundings—direct fragmentation processing could generate many irrelevant fragments, affecting storage space utilization and increasing processing costs. Therefore, the location corresponding to the first and / or second target can be determined based on the target image, and this location can be used as the target location, such as the local location of a defect on a panel. Then, the target image is cropped according to the target location, extracting a portion containing only the first and / or second target. This cropped portion can then be segmented. That is, in some embodiments, the segmentation of the target image includes: determining the target location in the target image; cropping the image according to the target location; and performing the segmentation process based on the cropped result.

[0090] In more specific application scenarios, cropping functions from relevant image processing libraries can be used to crop images according to the calculated cropping area. In related applications, functions such as `cv2.getRectSubPix` can be used for rectangular region cropping; or the `Image.crop` method can be used to crop, resulting in an image containing only the target (defect, etc.) portion. The resolution can be set to the default value, such as 512*512, and can also be dynamically adjusted based on the target information. Cropping is based on the width and height of the target information; if the width and height are greater than 512, the default values ​​are used; otherwise, the default values ​​are used. Afterwards, the cropped image is saved to the directory specified by `${source image disk} / ${date} / ${PanelId} / ${image list}`. A new filename can be generated based on the original image filename, target number, and other information to ensure image traceability and convenient management. Simultaneously, the appropriate image format, JPEG, can be selected for saving. Image quality parameters can be set according to requirements. The cropped images can be uniformly named using the rule: ${PanelId}_${Date}_${X}_${Y}_${Number}_${Type}.jpg. Further quality checks and verifications can be performed: the cropped images can be checked to see if the cropped area accurately contains the target (first or second target), and whether there are any omissions or extra parts. Verification can also be done through manual visual inspection or using simple image analysis algorithms (such as calculating the proportion of defective areas in the image) to ensure the accuracy of data cleaning.

[0091] In some embodiments, data unification can be performed before cropping the target image. In specific applications, coordinate calibration and unit unification can be performed on the image: check the coordinate and size units in the target information to ensure they are consistent with the image's pixel coordinate system. If there are unit differences, such as the target information being in millimeters while the image is in pixels, conversion must be performed according to a known scale or conversion relationship to ensure the coordinates and dimensions in the target information accurately correspond to the pixel positions in the image. Target location: based on the parsed and calibrated target position coordinate information, determine the center or boundary position of the target on the image. For example, if the target information provides the target center coordinates (x, y), then locate the corresponding (x, y) position in the image's pixel matrix; if it provides the coordinates of the top-left and bottom-right corners of the bounding box (x1, y1, x2, y2, x3, y3), then determine the bounding box's range.

[0092] In some embodiments, a target image corresponding to a first or second target may be represented by multiple target images. Therefore, to ensure data integrity, multiple target images corresponding to a first or second target are associated or grouped together. During storage, images of the same first or second target are retrieved in a loop. If the number of images increases, it indicates that there are still target images not associated with a group. If the number no longer increases, it is considered that all target images have been generated, and no new images will be generated. This allows storage based on multiple target images of a complete first or second target.

[0093] In some embodiments, since the target image or fragmented images need to be distributed for storage, the characteristics of blockchain can be well applied to this scenario. For example, blockchain's decentralized nature: blockchain nodes are formed using a distributed network, and images are fragmented and stored in a distributed manner. Immutability: defective image information and defect details are stored in the blockchain, ensuring the information cannot be tampered with, and any accidental operations on the image can be recovered. Encrypted digital signature: each fragment of the defective image is AES encrypted, and a hash operation is performed on the encrypted fragment to obtain a digital signature. Traceability: any operation on the defective image is traceable.

[0094] Subsequently, the network formed in the aforementioned embodiments can be configured using relevant blockchain concepts. Private blockchain networks, due to their superior security and privacy, can be considered an optimized network option. That is, in some embodiments, the network may include a private blockchain network. Taking the aforementioned scenario of panel production and testing as an example, using a private blockchain in a factory setting is particularly suitable because equipment is typically located within a local area network (LAN). Strict access control is essential, as factory production data and equipment management information are usually highly confidential and sensitive. A private blockchain can strictly restrict access to the blockchain network to only authorized nodes within the factory, ensuring data security and privacy. Furthermore, private blockchains are highly efficient and fast. Their consensus mechanism is relatively simple, and the number of nodes is limited, allowing for rapid consensus building within a LAN environment, improving transaction processing speed and meeting the needs of real-time data interaction and rapid response from factory equipment. Finally, private blockchains are less costly. Compared to public blockchains, private blockchains do not require significant computing power to maintain network security and run consensus algorithms. Building a private blockchain within a factory LAN can reduce hardware and energy costs. Customizability is key; factories can tailor their private blockchains to their specific business processes and needs, better adapting them to their production management and equipment control systems for more efficient business collaboration. Furthermore, the blockchain's consensus mechanism can utilize Raft, achieving strong consistency through leader election and log replication. This mechanism boasts low communication complexity, rapid consensus achievement, high real-time performance, and low resource consumption, making it suitable for factory scenarios.

[0095] Further, to further enhance the security of fragmented images. For any fragmented image, a corresponding image key can be generated randomly or according to a setting. The fragmented image is encrypted using the image key to obtain an encrypted image. Then, for the image key, when applied to the blockchain network, it can be encrypted according to the public key of the node storing the fragmented image to generate an encrypted key. Further, a hash calculation can be performed on the encrypted image, and then the hash calculation result is signed using the private key of the node storing the fragmented image, thereby completing the encryption and signature of the fragmented image in the blockchain and enhancing the security of the fragmented image. That is, in some embodiments, storing the at least two fragmented images in at least two nodes of the network includes: generating a corresponding image key for any fragmented image, encrypting the any fragmented image according to the image key to obtain an encrypted image; determining the public key and private key of the node storing the any encrypted image, encrypting the image key through the public key to obtain an encrypted key; performing a hash calculation on the any encrypted image, signing the hash calculation result according to the private key to obtain a signature result; storing the encrypted image, the encrypted key, and the signature result as the relevant data of the any fragmented image. In specific applications, this processing process can be performed on a certain local terminal device 101; it can also be performed on the server 102; it can also be partially performed on a certain local terminal device 101 and the other part on another local terminal device 101, etc.

[0096] In a more specific application scenario, for the fragmentation processing of a target image in a private chain, fragment segmentation: The Shamir secret sharing algorithm can be used to divide the target image into n fragments, and a threshold k (k < n) is set, that is, at least k fragments are required to restore the target image. For example, the image is divided into 5 fragments, and the threshold is set to 3, which means that the image can be restored only by collecting any 3 fragments. Each fragment has a unique identifier Fragment_ID, including information such as generation time and serial number. Then, metadata generation: Generate metadata for the image, including information such as the original name, format, size, hash value (which can be calculated through SHA-256, etc.), the number of fragments n, and the threshold k.

[0097] Next, encryption and digital signature are performed, specifically fragment encryption: A random AES (Advanced Encryption Standard) key (Key_i) is generated for each fragment image. This key is used to encrypt the fragment image, resulting in the encrypted fragment Encrypted_Fragment_i. Then, the RSA (Asymmetric Encryption Algorithm) public key of the receiving node is used to encrypt each AES key, resulting in Encrypted_Key_i. Here, for the encryption algorithm: AES (Advanced Encryption Standard) can be used for symmetric encryption, as it is mature, efficient, and can quickly encrypt image fragments; RSA can be used for asymmetric encryption, encrypting the symmetric encryption key to ensure secure key transmission; the hash function chosen is SHA-256, generating a unique hash value for integrity verification and digital signature. Digital signature: For each encrypted fragment Encrypted_Fragment_i, its hash value Hash_i is calculated using SHA-256. Then, the ECDSA private key of the node storing the fragment image is used to sign the hash value Hash_i, generating the digital signature Signature_i. Here, for digital signature algorithms: Elliptic Curve Digital Signature Algorithm (ECDSA) can be used, which, while ensuring security, requires less signature data and has higher computational efficiency compared to algorithms such as RSA.

[0098] Next, blockchain storage, node deployment, and network setup: Multiple blockchain node devices are deployed in a local area network environment to build a Hyperledger Fabric private blockchain network. Each node device is assigned a unique identity certificate, and strict access control ensures that only authorized nodes can participate in the network. Transaction construction: A transaction proposal is created, packaging the encrypted fragment (Encrypted_Fragment_i), encrypted key (Encrypted_Key_i), digital signature (Signature_i), fragment metadata, and overall image metadata into a transaction payload. The transaction proposal includes the operation type (e.g., storing image fragments), initiator identity information, etc. Consensus and storage: The initiator sends the transaction proposal to the endorsing nodes, which verify and endorse the proposal according to the smart contract (pre-deployed on the blockchain to verify the legality and integrity of the transaction). The endorsed proposal is sent to the ordering service nodes, which use the Raft consensus algorithm to order the transactions, generating ordered blocks. The blocks are broadcast to all nodes in the network. After verifying the legality of the transactions within the block, each node adds the block to its local blockchain, completing the storage of the image fragment information.

[0099] Furthermore, after storing the fragmented images, requests for acquiring or repairing the target image may be received. At this point, the system can first determine whether there is permission to acquire the fragmented images corresponding to the target image based on the acquisition request. For example, permission can be confirmed through the nodes storing these fragmented images, and the number of nodes that have passed permission confirmation can be reported. If the number of nodes that have passed permission confirmation exceeds a certain threshold (specifically, the threshold k set during image segmentation in the aforementioned embodiment), the acquisition request is considered acceptable. At this point, all fragmented images can be acquired and stitched together to form the target image; alternatively, the target image can be formed by repairing only the fragmented images whose permissions have been confirmed. In practical applications, for repair requests, it can be a process of repairing fragmented images stored by other nodes when an unexpected event occurs with the fragmented images of one or more nodes. The threshold is generally the aforementioned threshold k. When the number of fragmented images exceeds k, the target image can be restored using the corresponding repair mechanism.

[0100] In specific application scenarios, when accessing fragmented images for data collection, an access request can be submitted to the blockchain network. The network verifies the corresponding identity certificate and permission information to determine whether the user has the authority to obtain the corresponding fragmented image. Fragment Acquisition and Verification: If authorized, the encrypted fragment `Encrypted_Fragment_i`, the encrypted key `Encrypted_Key_i`, and the digital signature `Signature_i` are retrieved from the blockchain. The digital signature `Signature_i` is verified using the storage node's ECDSA public key. Simultaneously, the hash value of the encrypted fragment is recalculated using SHA-256 and compared with the corresponding hash value in the signature to verify the fragment's integrity and legitimate origin. Decryption and Image Restoration: `Encrypted_Key_i` is decrypted using the local RSA private key to obtain the original AES key `Key_i`. The encrypted fragment `Encrypted_Fragment_i` is then decrypted using the AES key `Key_i` to obtain the fragmented image. If at least k fragments can be collected, the original image can be restored using the Shamir secret sharing algorithm's restoration mechanism. That is, in some embodiments, after storing the at least two fragment images in at least two nodes of the network, the method further includes: in response to a request to obtain the target image, determining the permission to obtain the at least two fragment images for the request through the at least two nodes; in response to the ability to obtain more than a set threshold number of fragment images according to the access permission, forming the target image based on the at least two fragment images, thereby responding to the request to obtain the target image.

[0101] In a more specific application scenario, suppose there are 20 optical automated inspection (AOI) devices performing panel inspection, numbered AOI device 1, AOI device 2, AOI device 3... AOI device 20. Each AOI device is equipped with a data acquisition unit, and the data acquisition unit number is the same as the AOI device number. Assume AOI device 1 generates 6 defect images, represented as Defect_1.jpg, Defect_2.jpg... Defect_6.jpg.

[0102] Specific steps: (1) Fragment segmentation (single image processing). Input: The first defective image uploaded by data acquisition 1 (Defect_1.jpg). Operation: Use the Shamir secret sharing algorithm to segment the image into n=20 fragments (corresponding to 20 AOI devices), and set the threshold k=11 (at least 11 fragments can restore the image). Each fragment generates a unique identifier Fragment_ID (e.g., D1_F1_20250514 represents the first fragment of the first image). Output: 20 fragment files (e.g., D1_F1.bin-D1_F20.bin). Repeat operation: Perform the same operation on the 2nd to 6th images respectively, generating 20 fragments for each image, for a total of 6×20=120 fragments.

[0103] (2) Metadata generation (single image processing). The metadata content (taking the first image as an example) is generated as shown in Table 1.

[0104]

[0105]

[0106] Table 1. Metadata Content

[0107] The output is: each image generates an independent metadata file (Pic1_Metadata.json) containing the above information.

[0108] (3) Encryption and Digital Signature (Processing 120 Fragments One by One). Fragment Encryption (Single Fragment Processing): Input: The first fragment of the first image, D1_F1.bin. Operation: Randomly generate an AES key (Key_1), and encrypt D1_F1.bin using the AES algorithm to obtain the encrypted fragment Encrypted_D1_F1.bin. Encrypt Key_1 using the recipient's RSA public key (ADM system public key) to obtain Encrypted_Key_1. Output: Encrypted fragment Encrypted_D1_F1.bin and encryption key Encrypted_Key_1. Repeat this step: Encrypt 120 fragments one by one, generating an independent AES key for each fragment and encrypting it, generating a total of 120 sets of encrypted fragments + encryption keys. Digital Signature (Single Encrypted Fragment Processing): Input: Encrypted fragment Encrypted_D1_F1.bin. Operation: Calculate the hash value Hash_D1_F1 of the encrypted fragment using SHA-256. The target storage node (e.g., AOI device 1) signs Hash_D1_F1 using its own ECDSA private key, generating a digital signature Signature_D1_F1. Output: Digital signature Signature_D1_F1. Repeat this step: sign each of the 120 encrypted fragments one by one, generating 120 digital signatures.

[0109] (4) Blockchain Storage (120 fragments uploaded to the chain in batches). Node Deployment and Network Environment, Network Architecture: All 20 AOI devices are nodes in the blockchain network, including: Endorsing Nodes: AOI1, AOI2, AOI3 (responsible for verifying transactions). Sorting Nodes: AOI4, AOI5, AOI6 (Raft cluster responsible for block sorting). Ordinary Nodes: AOI7-AOI20 (synchronizing the ledger). Fragment Distribution and Storage (single image processing), taking the 20 fragments of the first image as an example: Transaction Construction: Send D1_F1.bin-D1_F20.bin to the corresponding AOI devices (e.g., send D1_Fi to AOIi). Each node (e.g., AOI1) constructs a transaction proposal, including: encrypted fragment D1_F1.bin, encrypted key Encrypted_Key_1, digital signature Signature_D1_F1, fragment metadata (Fragment_ID, hash value of the image to which it belongs), and overall image metadata (metadata of Defect_1.jpg). Consensus Process: Nodes send transaction proposals to endorsing nodes (AOI1, AOI2, AOI3). Endorsing nodes verify the proposal's legitimacy (e.g., permissions, hash consistency) and generate an endorsement signature. Ordering nodes (AOI4, AOI5, AOI6) collect the endorsed proposals, sort them according to the Raft algorithm, and generate blocks. All nodes verify and synchronize the blocks, writing the transactions to the ledger. Repeated Operation: The above process is executed one by one for each of the 120 fragments of the 6 images. Ultimately: On-chain storage: Metadata, encryption key, digital signature, and image hash value for each fragment. Off-chain storage: The actual encrypted fragment file (Di_Fi.bin) is stored on the local disk of the corresponding AOI device.

[0110] (5) Access and Verification (taking the first defective image as an example). Fragment Acquisition and Verification: Data Acquisition: Data acquisition queries the fragment distribution of the first image from the chain (e.g., D1_F1->AOI1, D1_F2->AOI2, ...). Requests the corresponding fragment's D1_F*.bin, Encrypted_Key_*, and Signature_D1_F* from at least 11 nodes (e.g., AOI1-AOI11). Integrity Verification: Verify Signature_D1_F* using the node's ECDSA public key to confirm the fragment's legitimate origin. Recalculate the SHA-256 hash value of D1_F*.i and compare it with the on-chain record to ensure it hasn't been tampered with. If someone deletes, modifies, or tampers with the fragment, repeat the above steps until all 11 fragments are found to be tampered with, and then redistribute and store the image in fragmented form. Decryption and Image Restoration: Key Decryption: Data acquisition uses its own RSA private key to decrypt the 11 Encrypted_Key_* fragments to obtain the original AES key. Fragment decryption: Decrypt 11 D1_F*.bin files using the AES key to obtain the original fragment. Image reconstruction: Merge the 11 fragments using the Shamir algorithm to reconstruct Defect_1.jpg, and compare and verify it with the image hash value stored on the blockchain.

[0111] Furthermore, performance optimizations can be implemented in different scenarios: (1) Batch transactions: Package 20 fragments of a single image into a single transaction proposal to reduce the number of on-chain transactions. (2) Fault recovery: If a node (such as AOI5) fails, the image can be recovered from the other 11 nodes without waiting for AOI5 to be repaired. The node status can be checked periodically, and the fragments on the failed node can be automatically redistributed.

[0112] In specific application scenarios, the formation of a blockchain network, since a blockchain is mainly composed of nodes, can be reflected by the addition of new nodes, thus demonstrating the current blockchain formation process. For example... Figure 3As shown, (1) New node starts. (2) Initialization: When a new node starts, it initializes the network configuration, including IP address, port number, etc. It initializes data storage and prepares to store the received data. (3) Construct broadcast message: The new node constructs a broadcast message containing its own network address and port number. (4) Send broadcast message: It uses the UDP protocol to send a broadcast message to the broadcast address (e.g., 255.255.255.255) in the local area network. (5) Other existing nodes in the network start. (6) Set up listener: When an existing node starts, it sets up a UDP listener to listen to the specified port (e.g., 8080). (7) Listen for broadcast messages: The listener continuously listens for broadcast messages from other nodes in the local area network. (8) Message processing: When the listener receives a broadcast message, it parses the message content and extracts the node's network address and port number. (9) Record new node information: The information of the new node is recorded in the node list for subsequent data synchronization and communication. For the node list: Each node maintains a node list to record the network information of all known nodes. The node list can be stored in memory or persisted to a file for recovery after a restart. Deduplication: When recording new node information, check if the same node information already exists in the node list to avoid duplicate recording. (10) Data synchronization, periodic synchronization: Each node periodically (e.g., every 10 seconds) requests the latest data from other nodes in the node list. Use HTTP / HTTPS or WebSocket for data transmission to ensure data consistency. On-demand synchronization: When a node generates new data, it actively notifies other nodes. The node receiving the notification requests the latest data from the node that sent the notification. (11) Data integrity verification, hash verification: Each data block is accompanied by a hash value to verify the integrity of the data. When a node receives data, it calculates the hash value of the data and compares it with the attached hash value to ensure data consistency. (12) Data storage.

[0113] In certain scenarios, version control can also be implemented: each data block is associated with a version number to identify the data version. Nodes compare version numbers when synchronizing data to ensure they obtain the latest data. Authentication: Digital certificates can be used to authenticate nodes, ensuring that only legitimate nodes can join the network. Messages can contain digital signatures to ensure message integrity and non-repudiation. Encrypted transmission: SSL / TLS protocols can be used to encrypt data transmission, preventing data from being stolen or tampered with during transmission.

[0114] In some embodiments, an object to be inspected that does not contain the first and second targets can be considered to have passed the inspection. To further improve inspection quality, a static image fine-tuning inspection can be performed on the object to be inspected. For example, during the inspection of a panel, the object to be inspected that has passed the aforementioned inspection can be sent to a high-resolution camera of a relevant optical automatic inspection device to take a picture of the panel. The image data acquisition component analyzes the static image in real time or uploads it to an analysis platform to identify static defects. If a defect is detected, it is immediately marked, triggering an alarm device to issue an alarm, and the panel enters the manual review stage.

[0115] As can be seen from the above embodiments, the present disclosure provides a target detection method. After determining the object to be inspected, the present disclosure first captures a video of the object to be inspected, obtaining a dynamic detection video. Then, it extracts a portion of frames from the detection video to form a detection image. This allows for dynamic detection of the detection video and static detection of the detection image, employing a dual-channel detection mode of video and image. The image, with its high resolution, clearly presents the subtle static structure of the object to be inspected; the video, on the other hand, records the dynamic process of the object to be inspected in real time, accurately capturing dynamic issues such as intermittent faults and displacement of moving parts. The two complement each other, comprehensively covering various targets, significantly improving detection accuracy, ensuring the integrity of the detection results, and providing a more reliable basis for product quality control.

[0116] It should be noted that the method of this disclosure embodiment can be executed by a single device, such as a computer or server. The method of this disclosure embodiment can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this disclosure embodiment, and the multiple devices will interact with each other to complete the method described.

[0117] It should be noted that the above description describes specific embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0118] Based on the same concept, corresponding to the methods of any of the above embodiments, this disclosure also provides a target detection device.

[0119] refer to Figure 4 The target detection device includes:

[0120] The first module 410 is used to track and capture images of the object to be inspected in response to the detection of the object to be inspected, thereby obtaining a detection video.

[0121] The second module 420 is used to perform a first target analysis based on the detected video.

[0122] The third module 430 is used to extract frame images from the detection video according to the set rules to obtain detection images.

[0123] The fourth module 440 is used to perform a second target analysis based on the detected image.

[0124] The fifth module 450 is configured to determine the object to be inspected as a target object in response to the presence of the first target and / or the second target.

[0125] In some exemplary embodiments, the second module 420 is further configured to:

[0126] In response to the presence of the first target, the time period in which the first target appears in the detected video is determined, and the image is cropped according to the time period to obtain at least one target image.

[0127] In some exemplary embodiments, the third module 430 is further configured to:

[0128] The duration of the detection video is determined, and the number and location of the detection images are determined based on the duration. The detection images are then extracted based on the number and location.

[0129] In some exemplary embodiments, the fifth module 450 is further configured to:

[0130] Determine the target image corresponding to the target object;

[0131] The target image is segmented to obtain at least two fragment images;

[0132] The at least two fragment images are stored in at least two nodes of the network; wherein the network comprises a plurality of nodes for data storage.

[0133] In some exemplary embodiments, the fifth module 450 is further configured to:

[0134] Determine the target location in the target image, and crop the image based on the target location;

[0135] The segmentation process is performed based on the truncation results.

[0136] In some exemplary embodiments, the fifth module 450 is further configured to:

[0137] Generate a corresponding image key for any fragment image, and encrypt the fragment image according to the image key to obtain an encrypted image;

[0138] Determine the public and private keys of the node storing any of the encrypted images, and encrypt the image key using the public key to obtain the encryption key;

[0139] Perform a hash calculation on any of the encrypted images, and sign the hash calculation result according to the private key to obtain a signature result;

[0140] The encrypted image, the encryption key, and the signature result are stored as relevant data for any fragment image.

[0141] In some exemplary embodiments, the network includes a private blockchain network.

[0142] In some exemplary embodiments, the fifth module 450 is further configured to:

[0143] In response to the request to obtain the target image, the at least two nodes determine the permission to obtain the at least two fragment images.

[0144] In response to the requirement that the number of fragment images can be obtained in greater than a set threshold according to the obtained permission, the target image is formed based on the at least two fragment images, thereby responding to the obtained request.

[0145] For ease of description, the above apparatus is described in terms of its functions, divided into various modules. Of course, in implementing the embodiments of this disclosure, the functions of each module can be implemented in one or more software and / or hardware.

[0146] The apparatus described above is used to implement the corresponding target detection methods in the foregoing embodiments and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0147] Based on the same concept, corresponding to the methods of any of the above embodiments, this disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the target detection method as described in any of the above embodiments.

[0148] Figure 5This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0149] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0150] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0151] The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components within the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touchscreens, microphones, various sensors, etc., while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0152] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0153] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0154] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0155] The electronic devices described above are used to implement the corresponding target detection methods in any of the foregoing embodiments and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0156] Based on the same concept, corresponding to the methods of any of the above embodiments, this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the target detection method as described in any of the above embodiments.

[0157] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0158] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the target detection method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0159] Based on the same concept, corresponding to the methods of any of the above embodiments, this disclosure also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processors to perform the target detection method. Corresponding to the execution entity for each step in each embodiment of the target detection method, the processor executing the corresponding step may belong to the corresponding execution entity.

[0160] The computer program products of the above embodiments are used to cause the computer and / or the processor to execute the target detection method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0161] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this disclosure (including the claims) is limited to these examples; within the framework of this disclosure, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this disclosure as described above, which are not provided in detail for the sake of brevity.

[0162] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this disclosure, the provided drawings may or may not show well-known power / ground connections to integrated circuit (IC) chips and other components. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this disclosure, and this also takes into account the fact that the details of implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this disclosure will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuitry) have been set forth to describe exemplary embodiments of this disclosure, it will be apparent to those skilled in the art that the embodiments of this disclosure may be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0163] Although this disclosure has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0164] This disclosure is intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A target detection method, characterized in that, include: In response to the detection of an object to be inspected, the object to be inspected is tracked and photographed to obtain a detection video; Perform first target analysis based on the detected video; Frame images are extracted from the detected video according to the set rules to obtain the detected images; Perform a second target analysis based on the detected image; In response to the presence of the first target and / or the second target, the object to be inspected is determined to be a target object.

2. The method according to claim 1, characterized in that, After performing the first target analysis based on the detected video, the method further includes: In response to the presence of the first target, the time period in which the first target appears in the detected video is determined, and image is cropped according to the time period to obtain at least one target image.

3. The method according to claim 1, characterized in that, The step of extracting frame images from the detected video according to a set rule to obtain the detected images includes: The duration of the detection video is determined, and the number and location of the detection images are determined based on the duration. The detection images are then extracted based on the number and location.

4. The method according to claim 1, characterized in that, After determining that the object to be inspected is the target object, the method further includes: Determine the target image corresponding to the target object; The target image is segmented to obtain at least two fragment images; The at least two fragment images are stored in at least two nodes of the network; wherein the network comprises a plurality of nodes for data storage.

5. The method according to claim 4, characterized in that, The segmentation process for the target image includes: Determine the target location in the target image, and crop the image based on the target location; The segmentation process is performed based on the truncation results.

6. The method according to claim 4, characterized in that, The step of storing the at least two fragment images in at least two nodes of the network includes: Generate a corresponding image key for any fragment image, and encrypt the fragment image according to the image key to obtain an encrypted image; Determine the public and private keys of the node storing any of the encrypted images, and encrypt the image key using the public key to obtain the encryption key; Perform a hash calculation on any of the encrypted images, and sign the hash calculation result according to the private key to obtain a signature result; The encrypted image, the encryption key, and the signature result are stored as relevant data for any fragment image.

7. The method according to claim 4, characterized in that, The network includes a private blockchain network.

8. The method according to claim 4, characterized in that, After storing the at least two fragment images in at least two nodes of the network, the method further includes: In response to the request to obtain the target image, the at least two nodes determine the permission to obtain the at least two fragment images. In response to the requirement that the number of fragment images can be obtained in greater than a set threshold according to the obtained permission, the target image is formed based on the at least two fragment images, thereby responding to the obtained request.

9. A target detection device, characterized in that, include: The first module is used to track and capture images of the object under inspection in response to the detection of the object under inspection, thereby obtaining a detection video. The second module is used to perform a first target analysis based on the detected video; The third module is used to extract frame images from the detected video according to the set rules to obtain the detected images; The fourth module is used to perform a second target analysis based on the detected image; The fifth module is used to determine the object to be inspected as a target object in response to the presence of the first target and / or the second target.

10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as claimed in any one of claims 1 to 8.

11. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions for causing a computer to perform the method as described in any one of claims 1 to 8.

12. A computer program product, characterized in that, It includes computer program instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1 to 8.

Images