Target tracking and identification method and device, and nonvolatile storage medium

Abstract

The application discloses a target tracking and identifying method and device and a nonvolatile storage medium. The method comprises the following steps: obtaining target image data of a target to be identified, wherein the target image data is image data of a preset part of the target to be identified; searching historical monitoring data included in a monitoring database, and taking the historical monitoring data meeting preset search conditions as initial search results; performing classification processing on the initial search results to obtain a plurality of classification results corresponding to the initial search results; performing parallel comparison processing on the plurality of classification results corresponding to the initial search results and the target image data to obtain a historical track corresponding to the target to be identified; and determining a target tracking and identifying result of the target to be identified in current monitoring data according to current monitoring data at a current moment and the historical track. The application solves the technical problems of low target tracking and identifying efficiency and low identifying accuracy in target monitoring and identifying in the related art.

Description

Technical Field

[0001] This invention relates to the field of target tracking and recognition, and more specifically, to a target tracking and recognition method, apparatus, and non-volatile storage medium. Background Technology

[0002] Currently, video surveillance equipment is widely distributed throughout daily life, playing a crucial role in bringing convenience and security. However, with the increasing number of surveillance devices, a significant investment of manpower and resources is required. Manual identification suffers from low accuracy in target detection and low search efficiency, making it difficult to achieve continuous identification and positioning. Due to the large amount of information in video images and other signals, related technologies often experience long processing times. In particular, efforts to improve recognition accuracy have increased the complexity of recognition algorithms, leading to poor recognition efficiency and low real-time performance.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a target tracking and identification method, apparatus, and non-volatile storage medium to at least solve the technical problems of low target tracking and identification efficiency and low identification accuracy in related technologies.

[0005] According to one aspect of the present invention, a target tracking and identification method is provided, comprising: acquiring target image data of a target to be identified, wherein the target image data is image data of a preset part of the target to be identified; retrieving historical monitoring data included in a monitoring database, and taking historical monitoring data that meets preset retrieval conditions as initial retrieval results, wherein the historical monitoring data is video and / or image data collected by a monitoring device; classifying the initial retrieval results to obtain multiple classification results corresponding to the initial retrieval results; performing parallel comparison processing on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain a historical trajectory corresponding to the target to be identified; and determining the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data at the current moment and the historical trajectory.

[0006] Optionally, when the search conditions include a search time range and a first similarity threshold, and the target image data includes face image data corresponding to the target to be identified, the step of searching the historical monitoring data included in the monitoring database and using the historical monitoring data that meets the preset search conditions as the initial search result includes: searching the historical monitoring data included in the monitoring database and using the historical monitoring data that meets the search time range as the first search data; determining the first similarity between the first search data and the face image data; and using the first search data whose first similarity is higher than the preset first similarity threshold as the initial search result.

[0007] Optionally, when the multiple classification results include face classification results, human body classification results, and vehicle body classification results, and the target image data includes face image data, human body image data, and vehicle body image data corresponding to the target to be identified, the parallel comparison processing of the multiple classification results corresponding to the initial retrieval result and the target image data to obtain the historical trajectory corresponding to the target to be identified includes: determining a first face feature vector corresponding to the target to be identified based on the face image data; determining a first human body feature vector corresponding to the target to be identified based on the human body image data; and determining a first vehicle body feature vector corresponding to the target to be identified based on the vehicle body image data; and determining a second human body feature vector corresponding to the face classification result. The system comprises: a face feature vector, a second human feature vector corresponding to the human body classification result, and a second vehicle feature vector corresponding to the vehicle body classification result; a first comparison process is performed on the first face feature vector and the second face feature vector to obtain face similarity; a second comparison process is performed on the first human feature vector and the second human feature vector to obtain human body similarity; and a third comparison process is performed on the first vehicle feature vector and the second vehicle feature vector to obtain vehicle body similarity, wherein the first comparison process, the second comparison process, and the third comparison process are performed in parallel; based on the face similarity, the human body similarity, and the vehicle body similarity, the historical trajectory corresponding to the target to be identified is obtained.

[0008] Optionally, obtaining the historical trajectory corresponding to the target to be identified based on the face similarity, the human body similarity, and the vehicle body similarity includes: taking historical monitoring data in the initial search results where the face similarity, the human body similarity, and the vehicle body similarity are greater than a preset second similarity threshold as the first search result; and obtaining the historical trajectory corresponding to the target to be identified based on the first search result.

[0009] Optionally, when the historical trajectory consists of multiple historical monitoring data included in the initial search result, determining the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data at the current moment and the historical trajectory includes: determining the historical monitoring data with the smallest time interval to the current moment among the multiple historical monitoring data corresponding to the historical trajectory, as the first monitoring data; determining whether the second similarity between the first monitoring data and the current monitoring data is greater than a preset third similarity threshold; if the second similarity between the first monitoring data and the current monitoring data is greater than the third similarity threshold, then using the first monitoring data as the initial identification result; and processing the data using a geocoding algorithm based on the initial identification result and a preset search distance threshold to determine the target tracking and identification result of the target to be identified in the current monitoring data.

[0010] Optionally, the step of processing the initial identification result and a preset search distance threshold using a geocoding algorithm to determine the target tracking and identification result of the target to be identified in the current monitoring data includes: processing the initial identification result as a search center point based on the search distance threshold using a geocoding algorithm to obtain candidate identification results; determining the candidate feature vector corresponding to the candidate identification result and the initial feature vector corresponding to the initial identification result; performing a fourth comparison processing between the candidate feature vector and the initial feature vector to obtain a third similarity; determining whether the third similarity is greater than a preset fourth similarity threshold; if the third similarity is greater than the fourth similarity threshold, then using the candidate identification result as the target tracking and identification result of the target to be identified in the current monitoring data.

[0011] Optionally, the method further includes: if the third similarity is not greater than the fourth similarity threshold, then obtaining the monitoring data corresponding to the next moment of the current moment, and updating the monitoring database with the current monitoring data as the historical monitoring data to obtain the updated monitoring database; taking the monitoring data corresponding to the next moment as the new current monitoring data, and performing the following operations repeatedly until the new third similarity is greater than the fourth similarity threshold: retrieving the new historical monitoring data included in the updated monitoring database, and taking the new historical monitoring data that meets the preset retrieval conditions as the new initial retrieval result, wherein the new historical monitoring data is the video and / or image data already collected by the monitoring device; classifying the new initial retrieval result to obtain the data corresponding to the new initial retrieval result. Multiple new classification results are obtained; the multiple new classification results corresponding to the new initial retrieval result and the target image data are compared in parallel to obtain a new historical trajectory corresponding to the target to be identified; a new initial identification result is obtained based on the new current monitoring data and the new historical trajectory; based on the search distance threshold, the new initial identification result is used as the new search center point, and the geocoding algorithm is used to process it to obtain a new candidate identification result; based on the new candidate feature vector corresponding to the new candidate identification result and the new initial feature vector corresponding to the new initial identification result, a new third similarity is obtained; if the new third similarity is greater than the fourth similarity threshold, the new candidate identification result is used as the target tracking identification result of the target to be identified in the new current monitoring data.

[0012] According to another aspect of the present invention, a target tracking and recognition device is provided, comprising: an acquisition module, configured to acquire target image data of a target to be identified, wherein the target image data is image data of a preset part of the target to be identified; a retrieval module, configured to retrieve historical monitoring data included in a monitoring database, and take historical monitoring data that meets preset retrieval conditions as initial retrieval results, wherein the historical monitoring data is video and / or image data collected by a monitoring device; a classification module, configured to classify the initial retrieval results to obtain multiple classification results corresponding to the initial retrieval results; a comparison module, configured to perform parallel comparison processing on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain a historical trajectory corresponding to the target to be identified; and a determination module, configured to determine the target tracking and recognition result of the target to be identified in the current monitoring data based on the current monitoring data at the current moment and the historical trajectory.

[0013] According to another aspect of the present invention, a non-volatile storage medium is provided, the non-volatile storage medium storing a plurality of instructions adapted for loading by a processor and executing any one of the target tracking and recognition methods described herein.

[0014] According to another aspect of the present invention, an electronic device is provided, comprising: one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement any one of the target tracking and recognition methods described herein.

[0015] In this embodiment of the invention, target image data of the target to be identified is acquired, wherein the target image data is image data of a preset part of the target to be identified; historical monitoring data included in the monitoring database is retrieved, and historical monitoring data that meets preset retrieval conditions is used as the initial retrieval result, wherein the historical monitoring data is video and / or image data collected by the monitoring equipment; the initial retrieval result is classified to obtain multiple classification results corresponding to the initial retrieval result; the multiple classification results corresponding to the initial retrieval result and the target image data are compared in parallel to obtain the historical trajectory corresponding to the target to be identified; based on the current monitoring data at the current moment and the historical trajectory, the target tracking and identification result of the target to be identified in the current monitoring data is determined. This achieves the goal of parallel image processing, thereby improving recognition efficiency, reducing recognition processing time, improving recognition efficiency, and improving the recognition accuracy of the target to be identified, thus solving the technical problems of low target tracking and identification efficiency and low recognition accuracy in related technologies for target monitoring and identification. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0017] Figure 1 This is a flowchart of an optional target tracking and identification method provided according to an embodiment of the present invention;

[0018] Figure 2 This is a schematic diagram of an optional target tracking and recognition method provided according to an embodiment of the present invention;

[0019] Figure 3 This is a schematic diagram of an optional target tracking and identification device provided according to an embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0023] Lightweight thread scheduling model (GMP model, Groutine & Channel model) is a scheduling model for multi-threaded high concurrency with parallel processing capabilities.

[0024] GeoHash algorithm is an algorithm used to locate latitude and longitude, which divides and encodes regions to improve positioning capabilities.

[0025] With video surveillance equipment widely deployed in every corner of life, these devices play a crucial role, bringing convenience and security. However, the increasing number of surveillance devices necessitates more manpower for constant monitoring, resulting in significant resource waste. Furthermore, manual target identification often suffers from low accuracy and efficiency, hindering continuous identification and location. Due to the large amount of information in video images and other signals, related technologies frequently experience long processing times and poor recognition efficiency.

[0026] To address the aforementioned problems, this invention provides a method embodiment for target tracking and identification. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0027] Figure 1 This is a flowchart of a target tracking and recognition method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0028] Step S102: Obtain target image data of the target to be identified, wherein the target image data is image data of a preset part of the target to be identified.

[0029] It is understandable that the first step is to determine the target to be identified, and use the target image data as the basis for identification. The target image data is the image data of a preset part of the target to be identified, that is, the associated target with the target to be identified, since it represents the target to be identified.

[0030] Optionally, there are multiple ways to obtain target image data. For example, when the target image data is set to be the image data of the target person, first upload the original image containing the target to be identified. The original image contains a crowd of people, including both the target person and the person to be identified. To find the target person in the crowd, a visual algorithm is used to identify and analyze the image data corresponding to each of the multiple people in the original image. The target person is then selected from the image data corresponding to each of the multiple people as the target to be identified, and the image data of the face portion is associated with it as the target image data of the target person.

[0031] In an optional embodiment, before obtaining the target image data of the target to be identified, the method further includes: obtaining security authentication information of the requesting account, wherein the requesting account is the account that requests to obtain the target image information of the target to be identified; and, if it is determined that the security authentication information has passed a preset security verification, obtaining a token corresponding to the requesting account, wherein the token indicates that the requesting account has passed the security verification, and the token is used to provide proof that the security verification has been passed when the requesting account interacts with information.

[0032] Understandably, the requesting account must first undergo security authentication, i.e., authorization verification. Only after confirming that the aforementioned security authentication information passes the preset security verification can the target image data be obtained. Furthermore, if access is granted, a token is issued to the requesting account. This token is used in computer authentication to avoid redundant account authentication processes. Any data interaction during the target tracking and identification process requires verification of this token as the basis for obtaining information and accessing computing power.

[0033] Step S104: Search the historical monitoring data included in the monitoring database, and take the historical monitoring data that meets the preset search conditions as the initial search results. The historical monitoring data is the video and / or image data that the monitoring equipment has collected.

[0034] It is understandable that preset search criteria are used as the basis for searching the monitoring database, and multiple historical monitoring data are used as the initial search results. By flexibly setting search criteria, the amount of data to be processed can be adjusted, which helps to control the balance between processing time and accuracy in the identification process.

[0035] In an optional embodiment, when the search conditions include a search time range and a first similarity threshold, and the target image data includes face image data corresponding to the target to be identified, the search of historical monitoring data included in the monitoring database, and the use of historical monitoring data that meets the preset search conditions as the initial search result, includes: searching the historical monitoring data included in the monitoring database, and using historical monitoring data that meets the search time range as the first search data; determining the first similarity between the first search data and the face image data; and using the first search data whose first similarity is higher than the preset first similarity threshold as the initial search result.

[0036] It's understandable that setting the search criteria to a search time range and a first similarity threshold is intended to achieve the goal of searching separately based on time and similarity. The initial search results consist of historical monitoring data that simultaneously meets the search time range and exceeds the first similarity threshold.

[0037] Step S106: Classify the initial search results to obtain multiple classification results corresponding to the initial search results.

[0038] It is understandable that the initial search results are historical monitoring data that meet the search criteria. In order to further identify and process the data, the initial search results will be classified to facilitate parallel comparison with the target image data.

[0039] Optionally, there are multiple classification processing methods described above. For ease of understanding, a specific example is given: when the target to be identified is a person, and the application scenario is finding a target person in traffic monitoring, the initial search results are classified into face results, body results, and vehicle results. The classification method can be adjusted based on specific changes in the target to be identified and the application scenario to obtain classification results with better recognition performance.

[0040] Step S108: Perform parallel comparison processing on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain the historical trajectory corresponding to the target to be identified.

[0041] It is understandable that, to improve the recognition accuracy of target image data, the target image data is compared with multiple classification results corresponding to the initial retrieval results in parallel. Since there are multiple classification results, the amount of data to be processed is large. To improve the recognition efficiency of target image data, multiple threads are invoked, demonstrating the ability to process multiple threads in parallel. This improves processing efficiency for the aforementioned multiple classification results, thus enhancing the overall recognition efficiency and ensuring the accuracy of the historical trajectory corresponding to the target to be identified.

[0042] In an optional embodiment, where the multiple classification results include face classification results, human body classification results, and vehicle body classification results, and the target image data includes face image data, human body image data, and vehicle body image data corresponding to the target to be identified, the parallel comparison processing of the multiple classification results corresponding to the initial retrieval result and the target image data to obtain the historical trajectory corresponding to the target to be identified includes: determining a first face feature vector corresponding to the target to be identified based on the face image data; determining a first human body feature vector corresponding to the target to be identified based on the human body image data; and determining a first vehicle body feature vector corresponding to the target to be identified based on the vehicle body image data; and determining the face classification results... The system takes the second face feature vector corresponding to the first face feature vector, the second human feature vector corresponding to the human body classification result, and the second vehicle feature vector corresponding to the vehicle body classification result; performs a first comparison process on the first face feature vector and the second face feature vector to obtain face similarity; performs a second comparison process on the first human feature vector and the second human feature vector to obtain human body similarity; and performs a third comparison process on the first vehicle feature vector and the second vehicle feature vector to obtain vehicle body similarity. The first, second, and third comparison processes are performed in parallel. Based on the face similarity, human body similarity, and vehicle body similarity, the historical trajectory corresponding to the target to be identified is obtained.

[0043] It can be understood that the initial search results correspond to multiple classification results, including face classification, human body classification, and vehicle body classification. These face, human, and vehicle body classification results are all obtained by classifying historical image data that meets the search criteria. The target image data serves as the basis for identifying the target. Given that the target image data is face image data, human body image data, or vehicle body image data, the face classification results are compared one-to-one with each other, the human classification results with each other, and the vehicle classification results with each other, using feature vectors to obtain face similarity, human body similarity, and vehicle body similarity. Based on these face similarity, human body similarity, and vehicle body similarity, the historical trajectory corresponding to the target to be identified is obtained.

[0044] Optionally, the GMP model can be used for parallel processing.

[0045] It should be noted that comparing and recognizing multiple classification results with target image data improves the accuracy of historical trajectories. Taking traffic monitoring as an example, for a driver's driving process, if the search conditions are set to the time range corresponding to the driving process, the driver's face, body, and vehicle should be collected by the monitoring equipment simultaneously. As is common sense, the face, body, and vehicle do not separate during driving. Therefore, using multiple face classification results, body classification results, and vehicle classification results for parallel comparison processing helps reduce the occurrence of misidentification and improve the accuracy of recognition.

[0046] In one optional embodiment, obtaining the historical trajectory corresponding to the target to be identified based on the face similarity, the human body similarity, and the vehicle body similarity includes: taking historical monitoring data in the initial search results where the face similarity, the human body similarity, and the vehicle body similarity are greater than a preset second similarity threshold as the first search result; and obtaining the historical trajectory corresponding to the target to be identified based on the first search result.

[0047] It is understandable that the similarity of faces, bodies, and vehicles is used to make judgments. If the similarity of faces, bodies, and vehicles is greater than the preset second similarity threshold, it is considered that the target to be identified has been identified. Based on the first search result, the historical trajectory of the target to be identified is obtained.

[0048] Optionally, if the face similarity, body similarity, and vehicle similarity are all greater than a preset second similarity threshold, the first search result is further restricted to obtain a first search result in which the face similarity, body similarity, and vehicle similarity are all greater than the second similarity threshold, and the historical trajectory is obtained.

[0049] Optionally, there are multiple ways to obtain historical trajectories, such as sorting the historical monitoring data in the first search result in reverse order and displaying the data points in groups.

[0050] It should be noted that, assuming clear images and no misjudgments during the recognition process, the driver's facial similarity, body similarity, and vehicle similarity should all be greater than a preset second similarity threshold. This means that facial, body, and vehicle features should be simultaneously recognizable within the same historical image data. However, in real-world applications, monitoring and data acquisition may not be ideal, as changes in lighting can affect the images, leading to misrecognition. To address these issues, while facial, body, and vehicle similarity should all exceed the second similarity threshold, historical image data from the historical trajectory can be obtained if one or more of these thresholds are greater than the threshold. This results in a large number of points, and since the target trajectory does not change abruptly, removing unreasonable historical image data with abrupt displacement changes can further improve the accuracy of the historical trajectory.

[0051] Step S110: Based on the current monitoring data at the current moment and the aforementioned historical trajectory, determine the target tracking and identification result of the target to be identified in the aforementioned current monitoring data.

[0052] It is understandable that by using the historical trajectory mentioned above, the movement process of the target to be identified in the historical image data is determined. Since the target to be identified does not have sudden displacement changes, determining the historical trajectory is helpful to eliminate falsely identified targets with sudden displacement changes. The target to be identified in the current monitoring data is determined by using the historical trajectory, which serves as the target tracking and identification result.

[0053] In an optional embodiment, when the historical trajectory consists of multiple historical monitoring data included in the initial retrieval result, determining the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data and the historical trajectory includes: determining the historical monitoring data with the smallest time interval to the current moment among the multiple historical monitoring data corresponding to the historical trajectory, as the first monitoring data; determining whether the second similarity between the first monitoring data and the current monitoring data is greater than a preset third similarity threshold; if the second similarity between the first monitoring data and the current monitoring data is greater than the third similarity threshold, then using the first monitoring data as the initial identification result; and based on the initial identification result and a preset search distance threshold, using a geocoding algorithm to process and determine the target tracking and identification result of the target to be identified in the current monitoring data.

[0054] It can be understood that the historical monitoring data with the smallest time interval to the current moment among multiple historical monitoring data corresponding to the historical trajectory is identified as the first monitoring data, i.e., the latest historical monitoring data. If the second similarity between the first monitoring data and the current monitoring data is greater than the third similarity threshold, it is considered that the target to be identified may have been found in the current monitoring data. To further confirm, based on the initial identification result greater than the third similarity threshold and the search distance threshold, a geocoding algorithm is used to process and determine the target tracking and identification result in the current monitoring data.

[0055] In an optional embodiment, the above-mentioned determination of the target tracking and identification result of the target to be identified in the current monitoring data by processing with a geocoding algorithm based on the initial identification result and a preset search distance threshold includes: processing with a geocoding algorithm based on the initial identification result as the search center point, using the initial identification result as the search distance threshold to obtain a candidate identification result; determining the candidate feature vector corresponding to the candidate identification result and the initial feature vector corresponding to the initial identification result; performing a fourth comparison processing on the candidate feature vector and the initial feature vector to obtain a third similarity; determining whether the third similarity is greater than a preset fourth similarity threshold; if the third similarity is greater than the fourth similarity threshold, then the candidate identification result is taken as the target tracking and identification result of the target to be identified in the current monitoring data.

[0056] It is understandable that, based on the principle that the displacement of the target to be identified does not change abruptly within a very short time interval, the distance difference between the target tracking and identification result and the initial identification result can be considered to be limited. Therefore, a search distance threshold is preset, and a geocoding algorithm is used for processing, with the initial identification result as the search center point. For candidate identification results within the above search range that have a third similarity greater than the preset fourth similarity threshold with the initial identification result, they are taken as target tracking and identification results.

[0057] In an optional embodiment, the method further includes: if the third similarity is not greater than the fourth similarity threshold, then acquiring the monitoring data corresponding to the next moment of the current moment, and updating the monitoring database with the current monitoring data as the historical monitoring data to obtain the updated monitoring database; using the monitoring data corresponding to the next moment as the new current monitoring data, and repeatedly performing the following operations until the new third similarity is greater than the fourth similarity threshold: retrieving the new historical monitoring data included in the updated monitoring database, and using the new historical monitoring data that meets the preset retrieval conditions as the new initial retrieval result, wherein the new historical monitoring data is video and / or image data already collected by the monitoring device; classifying the new initial retrieval result to obtain the new initial retrieval result. The system generates multiple new classification results; it performs parallel comparison processing on the multiple new classification results corresponding to the new initial retrieval results and the target image data to obtain a new historical trajectory corresponding to the target to be identified; based on the new current monitoring data and the new historical trajectory, it obtains a new initial identification result; based on the search distance threshold, it uses the new initial identification result as the new search center point and processes it using the geocoding algorithm to obtain a new candidate identification result; based on the new candidate feature vector corresponding to the new candidate identification result and the new initial feature vector corresponding to the new initial identification result, it obtains a new third similarity; if the new third similarity is greater than the fourth similarity threshold, it uses the new candidate identification result as the target tracking and identification result of the target to be identified in the new current monitoring data.

[0058] It is understandable that there may be situations where the target tracking and recognition result cannot be found in the current image data, requiring the target to be identified to be searched again in the next moment. Therefore, the current monitoring data at the current moment is stored in the monitoring database, and the retrieval continues based on the updated historical monitoring data. The monitoring data corresponding to the next moment is used as the new current monitoring data to perform the recognition search in a loop until the new third similarity is greater than the fourth similarity threshold, which is considered as finding the target tracking and recognition result in the new current monitoring data.

[0059] Through the above steps, image processing can be performed in parallel, thereby improving recognition efficiency. This achieves the technical effects of reducing recognition processing time, increasing recognition efficiency, and improving the recognition accuracy of the target to be identified. It also solves the technical problems of low target tracking and recognition efficiency and low recognition accuracy in related technologies for target monitoring and recognition.

[0060] Based on the above embodiments and optional embodiments, the present invention proposes an optional implementation method. Specific examples are provided for ease of understanding, such as: Figure 2This is a schematic diagram of an optional target tracking and recognition method provided according to an embodiment of the present invention, such as... Figure 2 As shown, in a specific application scenario set as traffic monitoring, where the target to be identified is a person, the steps to obtain the target tracking and identification result are as follows:

[0061] Step S1: Upload the original image containing the target person as the basis for identification. Use a visual algorithm to identify and analyze all targets contained in the original image. Among all the targets contained, there are at least multiple targets such as the target person and non-target people.

[0062] Step S2: Select the target person from the parsed original image, determine the target image data corresponding to the target person as the target person's face image data, and set the search conditions as the search time range and the first similarity threshold.

[0063] Step S3: Search historical monitoring data according to the search time range and the first similarity threshold to obtain historical monitoring data that meets the search conditions as the initial search results. Classify the obtained initial search results according to the target person's face, body, and vehicle body to obtain multiple classification results: face results, body results, and vehicle body results. To improve the recognition accuracy of the target image data, compare the face image data, body image data, and vehicle body image data with the multiple classification results corresponding to the initial search results in parallel. Since there are multiple classification results and the amount of data to be processed is large, a lightweight thread scheduling model is introduced to improve the recognition efficiency of the target image data. The lightweight thread scheduling model calls multiple threads and has the ability to process multiple threads in parallel, which can improve the processing efficiency for the face results, body results, and vehicle body results. Compare the face classification results with face image data, body classification results with body image data, and vehicle classification results with vehicle image data one by one using feature vectors to obtain face similarity, body similarity, and vehicle similarity. The first search result is obtained by judging based on facial similarity, human body similarity, and vehicle body similarity.

[0064] Step S4: If the facial similarity, human body similarity, and vehicle body similarity are all greater than the preset second similarity threshold, the target person is considered to have been identified. The historical image data in the first search result is sorted in reverse order and displayed in point groups to obtain the historical trajectory of the target person.

[0065] Step S5: Determine the historical monitoring data with the smallest time interval to the current moment from among multiple historical monitoring data corresponding to the historical trajectory, and use it as the first monitoring data, i.e., the latest historical monitoring data. If the second similarity between the first monitoring data and the current monitoring data is greater than the third similarity threshold, it is considered that the target person may have been found in the current monitoring data. To further confirm, based on the initial identification result greater than the third similarity threshold and the search distance threshold, a geocoding algorithm is used to process and determine the target tracking and identification result of the target person in the current monitoring data.

[0066] Based on the processing of steps S1 to S5 above, there are still target tracking and identification results that fail to find the target person, which will be explained in sub-step S51.

[0067] Step S51: Continue the target person identification search in the next moment. Therefore, store the current monitoring data in the monitoring database, and continue the search based on the updated historical monitoring data, i.e., return to step S3. Use the monitoring data corresponding to the next moment as the new current monitoring data to perform the identification search cyclically until a new third similarity is determined to be greater than the fourth similarity threshold in step S5, which is considered as finding the target tracking and identification result in the new current monitoring data.

[0068] The above-described optional implementation methods achieve at least one of the following effects: Compared to the method of directly determining the target to be identified in the current monitoring data in related technologies, which are limited by the monitoring and acquisition quality of the current monitoring data and may experience partial occlusion or light and shadow effects at the current moment, the accuracy of related technologies is greatly limited. Furthermore, the implementation method provided by this invention also applies a lightweight thread scheduling model for multi-threaded parallel identification processing, which can significantly improve identification efficiency, and a geocoding algorithm is applied in the search range processing.

[0069] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0070] This embodiment also provides a target tracking and recognition device for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the terms "module" and "device" can refer to a combination of software and / or hardware that performs a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementations, or a combination of software and hardware, are also possible and contemplated.

[0071] According to embodiments of the present invention, an apparatus embodiment for implementing a target tracking and identification method is also provided. Figure 3 This is a schematic diagram of a target tracking and identification device according to an embodiment of the present invention, such as... Figure 3 As shown, the target tracking and identification device includes: an acquisition module 302, a retrieval module 304, a classification module 306, a comparison module 308, and a determination module 310. The device will be described below.

[0072] The acquisition module 302 is used to acquire target image data of the target to be identified, wherein the target image data is image data of a preset part of the target to be identified;

[0073] The retrieval module 304, connected to the acquisition module 302, is used to retrieve historical monitoring data included in the monitoring database, and to use historical monitoring data that meets the preset retrieval conditions as the initial retrieval results. The aforementioned historical monitoring data is video and / or image data that has been collected by the monitoring equipment.

[0074] The classification module 306 is connected to the retrieval module 304 and is used to classify the initial retrieval results to obtain multiple classification results corresponding to the initial retrieval results.

[0075] The comparison module 308, connected to the classification module 306, is used to perform parallel comparison processing on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain the historical trajectory corresponding to the target to be identified.

[0076] The determination module 310, connected to the comparison module 308, is used to determine the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data at the current moment and the historical trajectory mentioned above.

[0077] In a target tracking and recognition device provided in this embodiment of the invention, an acquisition module 302 is set up to acquire target image data of a target to be identified, wherein the target image data is image data of a preset part of the target to be identified; a retrieval module 304, connected to the acquisition module 302, is used to retrieve historical monitoring data included in the monitoring database, and take the historical monitoring data that meets the preset retrieval conditions as the initial retrieval result, wherein the historical monitoring data is video and / or image data collected by the monitoring equipment; a classification module 306, connected to the retrieval module 304, is used to classify the initial retrieval result to obtain multiple classification results corresponding to the initial retrieval result; a comparison module 308, connected to the classification module 306, is used to perform parallel comparison processing on the multiple classification results corresponding to the initial retrieval result and the target image data to obtain the historical trajectory corresponding to the target to be identified; and a determination module 310, connected to the comparison module 308, is used to determine the target tracking and recognition result of the target to be identified in the current monitoring data based on the current monitoring data and the historical trajectory. It achieves the goal of parallel image processing, thereby improving recognition efficiency. It realizes the technical effects of reducing recognition processing time, improving recognition efficiency, and improving the recognition accuracy of the target to be identified. It also solves the technical problems of low target tracking and recognition efficiency and low recognition accuracy in related technologies.

[0078] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0079] It should be noted that the acquisition module 302, retrieval module 304, classification module 306, comparison module 308, and determination module 310 mentioned above correspond to steps S102 to S110 in the embodiments. The instances and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments. It should be noted that the above modules, as part of the device, can run on a computer terminal.

[0080] It should be noted that the optional or preferred implementation methods of this embodiment can be found in the relevant descriptions in the embodiments, and will not be repeated here.

[0081] The aforementioned target tracking and identification device may also include a processor and a memory. The acquisition module 302, retrieval module 304, classification module 306, comparison module 308, determination module 310, etc., are all stored as program units in the memory, and the processor executes the aforementioned program units stored in the memory to realize the corresponding functions.

[0082] The processor contains a core that retrieves the corresponding program unit from memory. One or more cores may be configured. Memory may include non-persistent memory in computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory includes at least one memory chip.

[0083] This invention provides a non-volatile storage medium storing a program that, when executed by a processor, implements a target tracking and recognition method.

[0084] This invention provides an electronic device, including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: acquiring target image data of a target to be identified, wherein the target image data is image data of a preset part of the target to be identified; retrieving historical monitoring data included in a monitoring database, and using historical monitoring data that meets preset retrieval conditions as initial retrieval results, wherein the historical monitoring data is video and / or image data already collected by the monitoring device; classifying the initial retrieval results to obtain multiple classification results corresponding to the initial retrieval results; performing parallel comparison processing on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain the historical trajectory corresponding to the target to be identified; and determining the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data and the historical trajectory. The device in this document can be a server, PC, etc.

[0085] This invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having the following method steps: acquiring target image data of a target to be identified, wherein the target image data is image data of a preset part of the target to be identified; retrieving historical monitoring data included in a monitoring database, and using historical monitoring data that meets preset retrieval conditions as initial retrieval results, wherein the historical monitoring data is video and / or image data already collected by the monitoring device; classifying the initial retrieval results to obtain multiple classification results corresponding to the initial retrieval results; performing parallel comparison processing on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain the historical trajectory corresponding to the target to be identified; and determining the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data and the historical trajectory.

[0086] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0087] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0088] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0090] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0091] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0092] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, 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-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0093] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0094] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A target tracking and recognition method, characterized in that, include: Acquire target image data of the target to be identified, wherein the target image data is image data of a preset part of the target to be identified; The historical monitoring data included in the monitoring database is retrieved, and the historical monitoring data that meets the preset search conditions is used as the initial search result. The historical monitoring data is video and / or image data that has been collected by the monitoring equipment. The initial search results are classified to obtain multiple classification results corresponding to the initial search results; Parallel comparison processing is performed on the multiple classification results corresponding to the initial retrieval results and the target image data to obtain the historical trajectory corresponding to the target to be identified; Based on the current monitoring data and the historical trajectory, determine the target tracking and identification result of the target to be identified in the current monitoring data; Where the historical trajectory is composed of multiple historical monitoring data included in the initial retrieval result, the step of determining the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data at the current moment and the historical trajectory includes: determining the historical monitoring data with the smallest time interval to the current moment among the multiple historical monitoring data corresponding to the historical trajectory, as the first monitoring data; determining whether the second similarity between the first monitoring data and the current monitoring data is greater than a preset third similarity threshold; if the second similarity between the first monitoring data and the current monitoring data is greater than the third similarity threshold, then taking the first monitoring data as the initial identification result; and based on the initial identification result and a preset search distance threshold, using a geocoding algorithm to process and determine the target tracking and identification result of the target to be identified in the current monitoring data. The step of determining the target tracking and identification result of the target to be identified in the current monitoring data by processing the initial identification result and a preset search distance threshold using a geocoding algorithm includes: processing the initial identification result as a search center point based on the search distance threshold to obtain candidate identification results; determining the candidate feature vector corresponding to the candidate identification result and the initial feature vector corresponding to the initial identification result; performing a fourth comparison processing between the candidate feature vector and the initial feature vector to obtain a third similarity; determining whether the third similarity is greater than a preset fourth similarity threshold; if the third similarity is greater than the fourth similarity threshold, then the candidate identification result is taken as the target tracking and identification result of the target to be identified in the current monitoring data.

2. The method according to claim 1, characterized in that, When the search conditions include a search time range and a first similarity threshold, and the target image data includes facial image data corresponding to the target to be identified, the step of searching the historical monitoring data included in the monitoring database, and using the historical monitoring data that meets the preset search conditions as the initial search results, includes: The historical monitoring data included in the monitoring database is retrieved, and the historical monitoring data that meets the retrieval time range is used as the first retrieval data; Determine the first similarity between the first retrieved data and the face image data; The first search data with a similarity higher than a preset first similarity threshold is used as the initial search result.

3. The method according to claim 1, characterized in that, When the multiple classification results include face classification results, human body classification results, and vehicle body classification results, and the target image data includes face image data, human body image data, and vehicle body image data corresponding to the target to be identified, the parallel comparison processing of the multiple classification results corresponding to the initial retrieval result and the target image data to obtain the historical trajectory corresponding to the target to be identified includes: Based on the face image data, a first face feature vector corresponding to the target to be identified is determined; based on the human body image data, a first human body feature vector corresponding to the target to be identified is determined; and based on the vehicle body image data, a first vehicle body feature vector corresponding to the target to be identified is determined. Determine the second face feature vector corresponding to the face classification result, the second human body feature vector corresponding to the human body classification result, and the second vehicle body feature vector corresponding to the vehicle body classification result; A first comparison process is performed on the first face feature vector and the second face feature vector to obtain face similarity; a second comparison process is performed on the first human body feature vector and the second human body feature vector to obtain human body similarity; and a third comparison process is performed on the first vehicle body feature vector and the second vehicle body feature vector to obtain vehicle body similarity, wherein the first comparison process, the second comparison process and the third comparison process are performed in parallel. Based on the facial similarity, the human body similarity, and the vehicle body similarity, the historical trajectory corresponding to the target to be identified is obtained.

4. The method according to claim 3, characterized in that, The step of obtaining the historical trajectory corresponding to the target to be identified based on the facial similarity, the human body similarity, and the vehicle body similarity includes: The historical monitoring data in the initial search results, including the face similarity, the human body similarity, and the vehicle body similarity, which are greater than a preset second similarity threshold, are used as the first search result. Based on the first search result, the historical trajectory corresponding to the target to be identified is obtained.

5. The method according to claim 1, characterized in that, The method further includes: If the third similarity is not greater than the fourth similarity threshold, then the monitoring data corresponding to the next moment of the current moment is obtained, and the current monitoring data is used as the historical monitoring data to update the monitoring database, so as to obtain the updated monitoring database; The monitoring data corresponding to the next moment is used as the new current monitoring data, and the following operations are performed repeatedly until the new third similarity is greater than the fourth similarity threshold: The updated monitoring database includes new historical monitoring data, which is then retrieved. New historical monitoring data meeting the preset retrieval conditions is used as the new initial retrieval result. This new historical monitoring data consists of video and / or image data collected by the monitoring equipment. The new initial retrieval result is then classified to obtain multiple new classification results. These multiple new classification results and the target image data are compared in parallel to obtain a new historical trajectory corresponding to the target to be identified. Based on the new current monitoring data and the new historical trajectory, a new initial identification result is obtained. Using the new initial identification result as the new search center point, and based on the search distance threshold, the geocoding algorithm is applied to obtain new candidate identification results. Finally, based on the new candidate feature vectors corresponding to the new candidate identification results and the new initial feature vectors corresponding to the new initial identification results, a new third similarity is obtained. If the new third similarity is greater than the fourth similarity threshold, the new candidate identification result is used as the target tracking and identification result of the target to be identified in the new current monitoring data.

6. A target tracking and identification device, characterized in that, include: The acquisition module is used to acquire target image data of the target to be identified, wherein the target image data is image data of a preset part of the target to be identified; The retrieval module is used to retrieve historical monitoring data included in the monitoring database, and to use historical monitoring data that meets the preset retrieval conditions as the initial retrieval results. The historical monitoring data is video and / or image data that has been collected by the monitoring equipment. The classification module is used to classify the initial search results to obtain multiple classification results corresponding to the initial search results; The comparison module is used to perform parallel comparison processing on the multiple classification results corresponding to the initial retrieval result and the target image data to obtain the historical trajectory corresponding to the target to be identified; The determination module is used to determine the target tracking and identification result of the target to be identified in the current monitoring data based on the current monitoring data at the current moment and the historical trajectory; The determining module is further configured to, when the historical trajectory is composed of multiple historical monitoring data included in the initial search result, determine the historical monitoring data with the smallest time interval to the current moment among the multiple historical monitoring data corresponding to the historical trajectory, and use it as the first monitoring data; determine whether the second similarity between the first monitoring data and the current monitoring data is greater than a preset third similarity threshold; if the second similarity between the first monitoring data and the current monitoring data is greater than the third similarity threshold, then use the first monitoring data as the initial identification result; based on the initial identification result and a preset search distance threshold, use a geocoding algorithm to process and determine the target tracking and identification result of the target to be identified in the current monitoring data; The determining module is further configured to: based on the search distance threshold, using the initial identification result as the search center point, process the data using a geocoding algorithm to obtain candidate identification results; determine the candidate feature vector corresponding to the candidate identification result and the initial feature vector corresponding to the initial identification result; perform a fourth comparison processing on the candidate feature vector and the initial feature vector to obtain a third similarity; determine whether the third similarity is greater than a preset fourth similarity threshold; if the third similarity is greater than the fourth similarity threshold, then use the candidate identification result as the target tracking and identification result of the target to be identified in the current monitoring data.

7. A non-volatile storage medium, characterized in that, The non-volatile storage medium stores multiple instructions, which are adapted to be loaded by a processor and executed by the target tracking and recognition method according to any one of claims 1 to 5.

8. An electronic device, characterized in that, include: One or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the target tracking and recognition method according to any one of claims 1 to 5.

Images