Method, apparatus, and storage medium for processing target objects based on swarm intelligence

Abstract

The present invention discloses a method, apparatus, and storage medium for processing target objects based on swarm intelligence. Among them, the method includes: detecting an input image within a first time period, and obtaining first attribute data of a target group from a target area displayed in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group; determining a target object from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior. The present invention solves the technical problem that in the existing pedestrian detection method, the analysis of pedestrian features ignores the interactive influence of the swarm intelligence evolution behavior of the surrounding crowd.

Description

Technical Field

[0001] The present invention relates to the field of detection technologies, and in particular, to a method, apparatus, and storage medium for processing target objects based on swarm intelligence. Background Art

[0002] In the pedestrian detection methods in the prior art, the analysis of pedestrian features is only based on the target pedestrian image sequence, that is, only for the trajectory of the target pedestrian. Its detection results are difficult to adapt to the changes in the flow patterns of different pedestrian volumes and different days (for example, holidays and weekdays), and false alarms are likely to occur. In addition, the above-mentioned existing technical solutions also require complex manual configuration to distinguish abnormal trajectories when applied in different locations.

[0003] In view of the above problems, no effective solution has been proposed yet. Summary of the Invention

[0004] Embodiments of the present invention provide a method, apparatus, and storage medium for processing target objects based on swarm intelligence, so as to at least solve the technical problem that in the pedestrian detection method in the prior art, the analysis of pedestrian features ignores the interactive influence of the swarm intelligence evolution behavior of the surrounding crowd.

[0005] According to one aspect of the embodiments of the present invention, there is provided a method for processing target objects based on swarm intelligence, including: detecting an input image in a first time period, and obtaining first attribute data of a target group from a target area shown in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group; determining a target object from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image in a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0006] According to another aspect of the embodiments of the present invention, there is also provided a method for processing a target object based on swarm intelligence, including: obtaining an input image within a first period; detecting the above input image, obtaining first attribute data of a target group from a target area shown in the above input image, and determining a target object from the above target group based on the above first attribute data and second attribute data, wherein the above first attribute data is used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group, the above second attribute data is obtained by detecting a historical image within a second period, the above second period is the previous period of the above first period, and the above target object has abnormal behavior; reporting a detection result to a server, wherein the above detection result is used to record the above target object with abnormal behavior.

[0007] According to another aspect of the embodiments of the present invention, there is also provided a method for processing a target object based on swarm intelligence, including: receiving an input image within a first period from a client; detecting the above input image, obtaining first attribute data of a target group from a target area shown in the above input image, and determining a target object from the above target group based on the above first attribute data and second attribute data, wherein the above first attribute data is used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group, the above second attribute data is obtained by detecting a historical image within a second period, the above second period is the previous period of the above first period, and the above target object has abnormal behavior; returning the detection result to the above client and displaying the above detection result on a graphical user interface of the above client, wherein the above detection result is used to record the above target object with abnormal behavior.

[0008] According to another aspect of the embodiments of the present invention, there is also provided a device for processing a target object based on swarm intelligence, including: a detection module, configured to detect an input image within a first period, and obtain first attribute data of a target group from a target area shown in the above input image, wherein the above first attribute data is used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group; a processing module, configured to determine a target object from the above target group based on the above first attribute data and second attribute data, wherein the above second attribute data is obtained by detecting a historical image within a second period, the above second period is the previous period of the above first period, and the above target object has abnormal behavior.

[0009] According to another aspect of the embodiments of the present invention, there is also provided a computer-readable storage medium. The computer-readable storage medium includes a stored program. When the program runs, it controls the device where the computer-readable storage medium is located to execute any one of the above-mentioned target object processing methods based on swarm intelligence.

[0010] According to another aspect of the embodiments of the present invention, there is also provided an electronic device, including: a processor; and a memory connected to the processor for providing instructions for the processor to perform the following processing steps: detecting an input image within a first time period, and obtaining first attribute data of a target group from a target area displayed in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group; determining a target object from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0011] In the embodiments of the present invention, an object detection method based on crowd modeling is adopted. By detecting an input image within a first time period, first attribute data of a target group is obtained from a target area displayed in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group; a target object is determined from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0012] It is easy to notice that in the embodiments of the present invention, by using input images in different time periods as detection targets, detecting the input image within the first time period, obtaining first attribute data of the target group from the target area displayed in the input image, and obtaining second attribute data by detecting the historical image within the second time period. Since the first attribute data and the second attribute data are input images in different time stages respectively, and in the second time period corresponding to the second attribute data, the target object in the collected historical image has abnormal behavior, therefore, the target group can be analyzed based on the first attribute data and the second attribute data, paying more attention to the interactive influence of the swarm intelligence evolution behavior of the surrounding people, and further the accuracy of determining the target object from the target group can be achieved.

[0013] Since the embodiments of the present invention pay more attention to the interactive influence of the group intelligence evolution behavior of the surrounding people of the target object, the purpose of improving the accuracy and detection efficiency of detecting the target object is achieved, thereby realizing the technical effect of more effectively performing security warning on the target object, and further solving the technical problem that the pedestrian detection method in the prior art ignores the interactive influence of the group intelligence evolution behavior of the surrounding people on the pedestrian features. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The drawings described herein are used to provide a further understanding of the present invention and form a part of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0015] Figure 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a method for processing a target object based on group intelligence;

[0016] Figure 2 is a flowchart of a method for processing a target object based on group intelligence according to an embodiment of the present invention;

[0017] Figure 3 is a regional plan view of a method for processing a target object based on group intelligence according to an embodiment of the present invention;

[0018] Figure 4 is an overall method flowchart of a method for processing a target object based on group intelligence according to an embodiment of the present invention;

[0019] Figure 5 is a flowchart of another method for processing a target object based on group intelligence according to an embodiment of the present invention;

[0020] Figure 6 is a flowchart of another method for processing a target object based on group intelligence according to an embodiment of the present invention;

[0021] Figure 7 is a schematic structural diagram of a device for processing a target object based on group intelligence according to an embodiment of the present invention;

[0022] Figure 8 is a structural block diagram of a computer terminal according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings 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 under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0025] First, some nouns or terms that appear during the description of the embodiments of the present invention are applicable to the following explanations:

[0026] Swarm Intelligence: Swarm Intelligence is a frontier method of machine intelligence that solves extremely large-scale complex problems by aggregating the wisdom of a group. It provides new technologies and means for solving extremely large-scale complex problems that are difficult to solve by traditional methods and has been widely applied in fields such as transportation, crowdsourcing computing, and software development. For specific application scenarios, domestic and foreign scholars have designed a variety of swarm intelligence methods from perspectives such as the construction of individual evaluation mechanisms, individual coding and decoding strategies, and group organizational structures.

[0027] Crowd Simulation: Crowd Simulation is the process of simulating the movement of a large number of entities or characters, which is usually used for crisis training, architecture and urban planning, and evacuation simulation, and can also be used for the creation of virtual scenes in movies or video games.

[0028] Object Detection: Object Detection is a branch of computer technology closely related to computer vision and image processing. Its goal is to detect specific semantic target entities in digital images and videos, such as people, buildings, cars, etc., and output the result as a rectangular box tightly enclosing the target entity. Object Detection has applications in many computer vision fields such as image retrieval and video acquisition.

[0029] Computer Vision: Computer vision is a science that studies how to enable machines to "see". More specifically, it refers to machine vision that uses cameras and computers to replace the human eye for tasks such as object recognition, tracking, and measurement, and further performs image processing to convert the images processed by the computer into images that are more suitable for human observation or transmission to instruments for detection. As a scientific discipline, computer vision studies related theories and technologies and attempts to build artificial intelligence systems that can obtain information from images or multi-dimensional data.

[0030] Embodiment 1

[0031] According to an embodiment of the present invention, an embodiment of target object processing based on swarm intelligence is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0032] The method embodiment provided by Embodiment 1 of the present invention can be executed on a mobile terminal, a computer terminal, or a similar computing device. Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing a method for processing target objects based on swarm intelligence is shown. As Figure 1 shown, the computer terminal 10 (or mobile device 10) may include one or more processors 102 (shown as 102a, 102b,..., 102n in the figure) (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions. In addition, it may further include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which can be included as one of the ports of the BUS bus), a network interface, a power supply, and / or a camera. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above-mentioned electronic device. For example, the computer terminal 10 may further include more or fewer components than Figure 1 shown, or have a different configuration from Figure 1 shown.

[0033] It should be noted that one or more of the above-mentioned processors 102 and / or other data processing circuits can generally be referred to as "data processing circuits" herein. The data processing circuit can be embodied in software, hardware, firmware, or any combination thereof, in whole or in part. In addition, the data processing circuit can be a single independent processing module, or be incorporated in whole or in part into any one of other elements in the computer terminal 10 (or mobile device). As involved in the embodiments of the present invention, the data processing circuit is a kind of processor control (such as the selection of a variable resistor terminal path connected to an interface).

[0034] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage devices corresponding to the method for processing a target object based on swarm intelligence in the embodiments of the present invention. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implements the above-mentioned method for processing a target object based on swarm intelligence. The memory 104 can include high-speed random access memory, and can also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory 104 can further include a memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 through a network. Examples of the above-mentioned network include but are not limited to the Internet, intranet, local area network, mobile communication network, and combinations thereof.

[0035] The transmission device 106 is used to receive or send data via a network. Specific examples of the above-mentioned network can include the wireless network provided by the communication provider of the computer terminal 10. In one instance, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station and thus communicate with the Internet. In one instance, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] The display can be, for example, a touch-screen liquid crystal display (LCD), which enables a user to interact with the user interface of the computer terminal 10 (or mobile device).

[0037] With the development of society and the increasing sophistication of the intercity transportation network, the population mobility in China is becoming greater and greater. On the one hand, it promotes the optimal utilization of labor resources. Therefore, currently, there is a great demand for video acquisition and detection of target objects from both the management perspective and the property security perspective. By means of supervisors' regular patrols and all-day video acquisition camera footage, although target objects can be identified in a timely and effective manner, this method places high requirements on the quantity and quality of supervisors and is difficult to popularize and deploy.

[0038] In recent years, with the development of computer vision technology, many intelligent target object detection methods have emerged. These methods generally compare the face data collected by face cameras with the information of target objects in the database to identify target objects. This method is limited by factors such as the deployment location, angle, and clarity of the cameras, and it is difficult to deploy on all cameras, resulting in a waste of video resources and the emergence of video acquisition blind spots. In addition, it cannot effectively warn of unknown target objects, and there are obvious deficiencies. There are also some methods that use object detection, pedestrian feature extraction, etc. to identify whether a person is suspicious by recognizing information such as the person's clothing, demeanor, and trajectory. Although this type of method avoids relying on face information, it still severs the behavioral connection between the individual and the whole. For example, pickpockets will tail and approach the pickpocket target, and there are often accomplices nearby to keep watch or assist. People involved in disputes and fights often cause passersby nearby to detour. Therefore, the overall behavioral analysis of the surrounding crowd helps to improve the accuracy of target object detection.

[0039] Under the above operating environment, the present invention provides a Figure 2 target object processing method based on swarm intelligence as shown. Figure 2 It is a flowchart of a target object processing method based on swarm intelligence according to an embodiment of the present invention. As Figure 2 shown, the method includes:

[0040] Step S202, detecting the input image within the first time period, and obtaining first attribute data of the target group from the target area shown in the above input image, where the above first attribute data is used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group;

[0041] Step S204, determining a target object from the above target group based on the above first attribute data and second attribute data, where the above second attribute data is obtained by detecting the historical image within the second time period, the above second time period is the previous time period of the above first time period, and the above target object has abnormal behavior.

[0042] As an alternative embodiment, the method for processing a target object based on swarm intelligence provided by the embodiments of the present invention can be, but is not limited to, applied to the field of urban brain projects. For example, it can be specifically implemented in front-end video capture devices, municipal target detection devices, etc.

[0043] Optionally, the above abnormal behaviors include, but are not limited to: pickpocketing, robbery, fighting, climbing over walls, etc., which are different from the behaviors performed by ordinary pedestrians during normal walking.

[0044] In the embodiments of the present invention, the above input image is Figure 3 the live video capture image obtained by a camera as shown, detect the live video capture image within the first time period, and obtain the first attribute data of the target group from the target area shown in the live video capture image; and, obtain the second attribute data after detecting the historical image within the second time period, where the second time period is the previous time period of the first time period, and the target object has abnormal behaviors during the second time period. Then, determine the target object from the target group based on the first attribute data and the second attribute data.

[0045] As an alternative embodiment, both the first attribute data and the second attribute data can be used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group.

[0046] Since the first attribute data and the second attribute data are input images in different time stages respectively, and during the second time period corresponding to the second attribute data, the target object in the collected historical image has abnormal behaviors, therefore, the target group can be analyzed based on the first attribute data and the second attribute data to determine the target object from the target group.

[0047] It is easy to notice that in the embodiments of the present invention, by using the input images in different time periods as the detection targets, detecting the input image within the first time period, obtaining the first attribute data of the target group from the target area shown in the input image, and obtaining the second attribute data after detecting the historical image within the second time period. Since the first attribute data and the second attribute data are input images in different time stages respectively, and during the second time period corresponding to the second attribute data, the target object in the collected historical image has abnormal behaviors, therefore, the target group can be analyzed based on the first attribute data and the second attribute data, paying more attention to the interactive influence of the swarm intelligence evolution behaviors of the surrounding people, and thus the accuracy of determining the target object from the target group can be achieved.

[0048] Since the embodiments of the present invention pay more attention to the interactive influence of the group intelligence evolution behavior of the surrounding people of the target object, the purpose of improving the accuracy and detection efficiency of detecting the target object is achieved, thereby realizing the technical effect of more effectively performing security warning on the target object, and further solving the technical problem that in the pedestrian detection method in the prior art, the analysis of pedestrian features ignores the interactive influence of the group intelligence evolution behavior of the surrounding people.

[0049] As an alternative embodiment, based on the above first attribute data and the above second attribute data, determining the target object from the above target group includes:

[0050] Step S302, predicting the flow trajectory of each target individual in the target group based on the second attribute data to obtain a first flow trajectory;

[0051] Step S304, determining the second flow trajectory of each target individual in the target group based on the first attribute data;

[0052] Step S306, comparing the second flow trajectory with the first flow trajectory to obtain a trajectory abnormality degree;

[0053] Step S308, determining the target object from the target group by using the trajectory abnormality degree.

[0054] Still as Figure 3 shown, the video capture live map and the pedestrian flow plane trajectory map obtained by the camera can be displayed in the form of a regional plane map. Among them, according to the pre-calibrated spatial relationship, the pedestrian flow trajectory and position information obtained by recognizing the video capture live map are mapped onto the plane map to obtain the pedestrian flow plane trajectory map. The pedestrian flow plane trajectory map is used to represent the pedestrian flow detection and tracking results of the target group in the input image and serves as the basic input of the group intelligence pedestrian flow model; detection frames of abnormal behaviors such as pickpocketing, fighting, and climbing over the wall captured by the group intelligence pedestrian flow model are drawn on the video capture live map, and the analysis results output by the group intelligence pedestrian flow model are received, that is, the matching results of the flow direction of the people in the area after the group intelligence algorithm is modeled. For example, the analysis result can be: there is or there is no target object with abnormal behavior.

[0055] As an alternative embodiment, the above group intelligence pedestrian flow model can determine the second flow trajectory of each target individual in the target group based on the first attribute data, and predict the flow trajectory of each target individual in the target group based on the second attribute data to obtain a first flow trajectory; after determining the first flow trajectory and the second flow trajectory, the trajectory abnormality degree between the first flow trajectory and the second flow trajectory can be calculated by comparison, and then the target object can be determined from the target group by using the trajectory abnormality degree.

[0056] Optionally, it is determined that the above second attribute data can be completed before going online, and it is determined that the above first attribute data can be continuously executed at an interval of the first time interval (small, such as 80 ms) after going online, and the swarm intelligence trajectory analysis is periodically executed online after the second time interval (large, such as 5 min).

[0057] As another optional embodiment, after determining the first flow trajectory and the second flow trajectory, the first flow trajectory and the second flow trajectory can also be marked with different colors on the regional floor plan.

[0058] As an optional embodiment, determining the target object from the above target group by using the above trajectory anomaly degree includes: when the above trajectory anomaly degree is greater than the first threshold, determining the target individual corresponding to the above trajectory anomaly degree as the above target object.

[0059] Through the above optional embodiments, the solution of the present invention does not depend on face information and the target object library, but combines the swarm intelligence crowd flow model and the computer vision method. On the one hand, each target individual in the target group is detected by the vision method to identify abnormal behaviors existing in the target group. On the other hand, the overall crowd behavior is modeled and simulated by the swarm intelligence algorithm, and the abnormal personnel are analyzed from the perspective of intelligent group interaction to comprehensively determine the target object.

[0060] As an optional embodiment, predicting the flow trajectory of each target individual in the above target group based on the above second attribute data to obtain the above first flow trajectory includes:

[0061] Step S402, analyzing the above second attribute data by using a swarm intelligence evolution model to obtain the above first flow trajectory, wherein the swarm intelligence evolution model is trained by using the historical attribute data in the above target area, and the swarm intelligence evolution model is used to predict the flow trajectory of each target individual in the above target group.

[0062] Optionally, the above swarm intelligence evolution model is a crowd flow prediction model. As an optional embodiment, the historical attribute data of the target area, for example, historical crowd flow data, can be used to model and train the crowd flow prediction model based on the swarm intelligence method to obtain the crowd flow prediction model. The crowd flow prediction model is used to estimate the normal trajectory of pedestrians and compare it with their actual trajectories to analyze the suspicious degree of the trajectories.

[0063] For example, the target object video data of the public data set or the swarm intelligence database can be used, but not limited to, to extract the high-dimensional feature vectors of each pedestrian (representing information such as the clothing, expression, and gait of the pedestrian).

[0064] Optionally, in the embodiments of the present invention, modeling the flow of people is related to the spatial environment of the target area (for example, wall obstacles, entrance and exit directions, etc.). Therefore, historical data in the target area needs to be used for modeling and optimization; learning the visual features (such as clothing, expression, gait, etc.) of the target objects that appear in the second time period. If the correlation with the regional environment is relatively low, an external data set can also be used for learning.

[0065] Through the embodiments of the present invention, a method for crowd modeling based on computer vision combined with swarm intelligence detects target objects among pedestrians through their behavior patterns, avoiding the dependence on face cameras and face databases in the prior art, reducing the usage threshold, and being more capable of essentially discovering and alarming target objects, with stronger effectiveness and versatility.

[0066] In the prior art, there are technical solutions for analyzing pedestrian trajectories, but they only target the trajectories of target pedestrians and ignore the mutual influence among the pedestrian crowd. The results are difficult to adapt to the changes in the flow patterns of different pedestrian volumes and different days (such as holidays and weekdays), resulting in false alarms. The solution of the present invention models the flow of people based on the swarm intelligence method, and can estimate the behavior patterns of the crowd with different pedestrian volumes and on different days with self - adaptability and higher accuracy, thereby improving the accuracy of detecting target objects.

[0067] As an optional embodiment, the method for processing target objects based on swarm intelligence provided by the embodiments of the present invention proposes a method for detecting target objects based on crowd modeling. By identifying and structurally recording the positions, trajectories, and personal characteristics of the crowd in the target area, the detection method based on swarm intelligence models the flow behavior of the crowd. By comparing the real - time trajectories of people with the simulation prediction results, people with abnormal trajectories are identified. In addition, combined with computer vision methods, it is determined whether target objects have abnormal behaviors such as pickpocketing, fighting, and climbing over the wall, comprehensively detecting target objects in the video acquisition area and reporting them. This solution avoids the identification limitations of the traditional method of identifying target objects by comparing with a face database for unknown people and the dependence on face detection cameras, and by combining the swarm intelligence modeling method, expands the identification dimension of potential target objects, and can more effectively detect and alarm target objects.

[0068] As an optional embodiment, the above - mentioned method for processing target objects based on swarm intelligence further includes:

[0069] Step S602: Detect the above - mentioned input image, and obtain the feature data of the above - mentioned target group from the above - mentioned target area, where the feature data is used to describe the visual features of each target individual in the above - mentioned target group.

[0070] Step S604, when the above-mentioned trajectory abnormality degree is less than or equal to the first threshold, analyze the above-mentioned first attribute data and the above-mentioned feature data by using the target neural network model to obtain the confidence level of each target individual in the above-mentioned target group, where the above-mentioned target neural network model is trained by using the associated data set of the above-mentioned target object, and the above-mentioned target neural network model is used to calculate the confidence level of whether there is abnormal behavior for each target individual in the above-mentioned target group;

[0071] Step S606, when the above-mentioned confidence level is greater than the second threshold, determine the target individual corresponding to the above-mentioned confidence level as the above-mentioned target object.

[0072] Optionally, the above-mentioned associated data set is the target object video data of the public data set or the crowd intelligence database; the above-mentioned target neural network model is trained by using the associated data set of the above-mentioned target object, and the above-mentioned target neural network model is used to calculate the confidence level of whether there is abnormal behavior for each target individual in the above-mentioned target group. For example, the above-mentioned target neural network model can be a classification network model, and this classification network model is used to analyze the suspicious degree according to the high-dimensional features of pedestrians.

[0073] As an optional embodiment, the above-mentioned second threshold is set based on the above-mentioned trajectory abnormality degree.

[0074] As an optional embodiment, as Figure 4 shown in the overall method flow chart, in the embodiment of the present invention, the input image can also be detected to identify the first attribute data of the target group in the target area shown in the input image. For example, the position and quantity of the target group in the input image are identified by the target detection and head counting methods, and the feature data of the above-mentioned target group is obtained from the above-mentioned target area, that is, the high-dimensional feature vector of pedestrians is extracted.

[0075] In the embodiment of the present invention, a crowd flow prediction model is trained by using the historical attribute data in the target area, and by adding a head counting method on the basis of target detection, the number of pedestrians in the area can be estimated more accurately, ensuring the accuracy of the crowd flow prediction model; still as Figure 4 shown, when the above-mentioned trajectory abnormality degree is less than or equal to the first threshold, a classification network model is used to analyze the above-mentioned first attribute data and the above-mentioned feature data to obtain the confidence level of each target individual in the above-mentioned target group. For example, an action recognition method can also be used to identify whether there are abnormal behaviors such as pickpocketing, fighting, and climbing over the wall by the people in the picture. If the confidence level of each target individual in the above-mentioned target group is obtained and it is determined that the above-mentioned confidence level is greater than the second threshold, then the target individual corresponding to the above-mentioned confidence level is determined as the above-mentioned target object and an alarm is directly issued.

[0076] Still as Figure 4As shown, structured data of crowd trajectories is extracted through a target tracking method in combination with historical data and stored in a storage queue (for example, a crowd simulation queue); then, through swarm intelligence trajectory analysis, the pedestrian flow data of the previous time period is input into the above-mentioned pedestrian flow prediction model to predict the crowd flow direction of the current time period, and a prediction result is obtained. If the above storage queue reaches the set duration, the flow trajectory of each pedestrian in the first time period is compared with the above prediction result to obtain a trajectory anomaly degree. If the above trajectory anomaly degree is greater than the first threshold, the target individual corresponding to the confidence level is determined as the target object and an alarm is issued; otherwise, the second threshold is calculated and output through a preset rule.

[0077] Optionally, the above trajectory anomaly degree is a score between 0 and 1. The above preset rule is to set the second threshold according to the trajectory anomaly degree. When the trajectory anomaly degree is relatively high, a lower second threshold is set, and vice versa. Specifically, the preset rule can be continuous. For example, the second threshold = max(0.3, 0.9 - trajectory anomaly degree), or the preset rule can also be segmented. For example, if the trajectory anomaly degree is in [0.9, 1], the second threshold is 0.5; if the trajectory anomaly degree is in [0.7, 0.9), the second threshold is 0.7; if the trajectory anomaly degree is in [0.5, 0.7), the second threshold is 0.8; if the trajectory anomaly degree is in [0, 0.5), the second threshold is 0.9. For another example, when the trajectory anomaly degree is greater than 0.9, the second threshold can be set to 0.5, and when the trajectory anomaly degree is greater than 0.7, the second threshold can be set to 0.7; for another example, when the trajectory anomaly degree is greater than 0.5, the second threshold can be set to 0.8, otherwise it is set to 0.9.

[0078] As an optional embodiment, when the storage queue reaches the set duration in the previous cycle, calculations are started and executed in parallel with the real-time detection process to avoid interfering with the real-time execution of the real-time detection process; multiple possible predicted trajectories can be output for the same pedestrian; and in the above optional embodiment, the actual trajectory and the predicted trajectory of the pedestrian within the time period can also be compared from dimensions such as the walking speed and detour mode of the target individual, and the trajectory anomaly degree is obtained through a preset rule.

[0079] As an optional embodiment, the above-mentioned target object processing method based on swarm intelligence further includes:

[0080] Step S702: Perform action recognition on the above input image, and obtain the confidence level of whether there is abnormal behavior for each target individual in the above target group from the above target area;

[0081] Step S704: When the above confidence level is greater than the third threshold, determine the target individual corresponding to the above confidence level as the above target object.

[0082] Optionally, by performing action recognition on the above input image, the above pedestrian high-dimensional feature vector and the action recognition result can be combined into a feature vector, which is input into a pre-trained target object classification model, and the confidence of each target individual in the target group within the target area calculated by the target object classification model is received. If the confidence is greater than a third threshold, the target individual corresponding to the confidence is determined as a suspicious object and an alarm is issued.

[0083] As an alternative embodiment, instead of using the action recognition method to detect abnormal behaviors, a target detection method can be used. The human body images performing pickpocketing, fighting and other behaviors are used as detection targets, and whether abnormal behaviors occur is determined based on the detection results of adjacent multiple frames. Modifying to use the target detection method for visual recognition of abnormal behaviors can avoid the requirement for video sequences in the action recognition method and reduce cache occupancy. However, in some cases, the determination accuracy may be lower than that of the action recognition method.

[0084] As another alternative embodiment, if conditions permit, in the embodiments of the present invention, a face camera can be further combined to perform behavior recognition on the target individual. By comparing the identity and behavior records and other information recorded in the public data set or the swarm intelligence database, or by counting the frequency of its appearance in the monitoring area within a certain period of time (such as one week), it is also used as a high-dimensional feature and used as network input to calculate the confidence of the target object, further improving the recognition effect of the target object.

[0085] It should be noted that the above-mentioned target detection, head counting, pedestrian high-dimensional feature extraction, action recognition method and target tracking method of the present invention can all be performed through a pre-trained deep neural network model. The above-mentioned deep neural network model is relatively common and can be directly implemented using publicly available models and pre-trained weights provided by other methods, or can be re-modeled and trained by itself using publicly available data sets or constructing data sets to obtain more accurate results, or can also be implemented using numerical optimization methods that do not require pre-training.

[0086] It can be seen from the solution of the present invention that based on computer vision methods such as target detection and action recognition, combined with the crowd modeling method of swarm intelligence, on the one hand, starting from action recognition, abnormal behaviors such as pickpocketing, fighting, and climbing over the wall by personnel are recognized, and on the other hand, target objects with relatively suspicious trajectories are recognized from the flow trajectory pattern, comprehensively determining and recognizing the target object, and positioning and alarming it. For example, in the embodiments of the present invention, by combining the swarm intelligence crowd modeling method to collect the normal crowd behavior patterns of the landing scene, the swarm intelligence behavior patterns of the regional crowd can be modeled, so as to adaptively recognize the target objects among them.

[0087] For another example, with the assistance of crowd modeling in the embodiments of the present invention, problems such as the inability to recognize minor actions such as pickpocketing due to limitations in camera position and resolution when using only computer vision methods can be solved. Additionally, by modeling based on swarm intelligence methods, the interactions between individual pedestrians can be estimated more accurately, the flow patterns of pedestrians under different pedestrian volumes and different days (such as holidays and weekdays) can be adaptively estimated, and the accuracy of detecting suspicious pedestrian flows can also be significantly improved.

[0088] Under the above operating environment, the present invention provides a method for processing target objects based on swarm intelligence as Figure 5 shown. Figure 5 FIG. is a flowchart of another method for processing target objects based on swarm intelligence according to an embodiment of the present invention. As Figure 5 shown, the method includes:

[0089] Step S802, obtaining an input image within a first time period;

[0090] Step S804, detecting the above input image, obtaining first attribute data of a target group from the target area shown in the above input image, and determining a target object from the above target group based on the above first attribute data and second attribute data. Among them, the above first attribute data is used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group, the above second attribute data is obtained by detecting historical images within a second time period, the above second time period is the previous time period of the above first time period, and the above target object has abnormal behavior;

[0091] Step S806, reporting the detection result to the server, where the above detection result is used to record the above target object with abnormal behavior.

[0092] Optionally, the above execution entity is a SaaS client. By means of communication and data interaction between the SaaS client and the SaaS server, the SaaS client is used to obtain an input image within a first time period, detect the above input image, obtain first attribute data of a target group from the target area shown in the above input image, and determine a target object from the above target group based on the above first attribute data and second attribute data, and then report the detection result to the server, so that the above target object with abnormal behavior can be recorded in the server.

[0093] As an optional embodiment, the method for processing target objects based on swarm intelligence provided by the embodiments of the present invention can be, but is not limited to, applied to the field of urban brain projects. For example, it can be specifically implemented in front-end video acquisition devices and municipal target detection devices, etc.

[0094] Optionally, the above abnormal behaviors include, but are not limited to: pickpocketing, robbery, fighting, climbing over the wall, etc., which are different from the behaviors performed by ordinary pedestrians during normal walking.

[0095] In the embodiment of the present invention, the above input image is a live video capture image obtained by a camera as shown in Figure 3 Detect the live video capture image in the first time period, and obtain the first attribute data of the target group from the target area shown in the above live video capture image; and, obtain the second attribute data after detecting the historical image in the second time period. The second time period is the previous time period of the first time period, and the target object had abnormal behaviors during this second time period. Then, determine the target object from the above target group based on the above first attribute data and second attribute data.

[0096] As an optional embodiment, both the above first attribute data and the above second attribute data can be used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group.

[0097] Since the above first attribute data and second attribute data are input images in different time stages respectively, and during the second time period corresponding to the second attribute data, the target object in the collected historical images had abnormal behaviors, therefore, the above target group can be analyzed based on the first attribute data and second attribute data to determine the target object from the above target group.

[0098] It is easy to notice that in the embodiment of the present invention, by using the input images in different time periods as the detection targets, detecting the input image in the first time period, obtaining the first attribute data of the target group from the target area shown in the above input image, and obtaining the second attribute data after detecting the historical image in the second time period. Since the above first attribute data and second attribute data are input images in different time stages respectively, and during the second time period corresponding to the second attribute data, the target object in the collected historical images had abnormal behaviors, therefore, the above target group can be analyzed based on the first attribute data and second attribute data, paying more attention to the interactive influence of the group intelligence evolution behaviors of the surrounding people, and then the accuracy of determining the target object from the above target group can be achieved.

[0099] Since the embodiment of the present invention pays more attention to the interactive influence of the group intelligence evolution behaviors of the people around the target object, it achieves the purpose of improving the accuracy and detection efficiency of detecting the target object, thus realizing the technical effect of more effectively performing security warning on the target object, and further solving the technical problem that the existing pedestrian detection method ignores the interactive influence of the group intelligence evolution behaviors of the surrounding people in the analysis of pedestrian characteristics.

[0100] Under the above operating environment, the present invention provides a method for processing target objects based on swarm intelligence as Figure 6 shown. Figure 6 It is a flowchart of another method for processing target objects based on swarm intelligence according to an embodiment of the present invention. As Figure 6 shown, the method includes:

[0101] Step S902, receiving an input image within a first time period from a client;

[0102] Step S904, detecting the above input image, obtaining first attribute data of a target group from a target area shown in the above input image, and determining a target object from the above target group based on the above first attribute data and second attribute data. Among them, the above first attribute data is used to describe the position of each target individual in the above target group in the above input image and the number of individuals included in the above target group, the above second attribute data is obtained by detecting a historical image within a second time period, the above second time period is the previous time period of the above first time period, and the above target object has an abnormal behavior;

[0103] Step S906, returning a detection result to the above client and displaying the detection result on a graphical user interface of the above client, where the above detection result is used to record the above target object with an abnormal behavior.

[0104] Optionally, the above execution entity is a SaaS server. By means of communication and data interaction between a SaaS client and the SaaS server, the SaaS client is used to obtain an input image within a first time period, detect the above input image, obtain first attribute data of a target group from a target area shown in the above input image, and determine a target object from the above target group based on the above first attribute data and second attribute data. The above target object with an abnormal behavior is recorded in the server, and a detection result is returned to the above client and displayed on a graphical user interface of the above client.

[0105] As an optional embodiment, the method for processing target objects based on swarm intelligence provided by the embodiments of the present invention can be, but is not limited to, applied to the field of urban brain projects. For example, it can be specifically implemented in front-end video acquisition devices and municipal target detection devices, etc.

[0106] Optionally, the above abnormal behaviors include, but are not limited to: pickpocketing, robbery, fighting, climbing over a wall, etc., which are different from the behaviors performed by ordinary pedestrians during normal walking.

[0107] In the embodiments of the present invention, the above input image is as Figure 3The live video capture image obtained by the camera is shown. The live video capture image in the first time period is detected, and the first attribute data of the target group is obtained from the target area shown in the above live video capture image; and, the second attribute data is obtained after detecting the historical image in the second time period. The second time period is the previous time period of the first time period, and the target object had abnormal behavior in the second time period. Then, based on the first attribute data and the second attribute data, the target object is determined from the target group.

[0108] As an optional embodiment, both the first attribute data and the second attribute data can be used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group.

[0109] Since the first attribute data and the second attribute data are input images in different time stages respectively, and in the second time period corresponding to the second attribute data, the target object in the collected historical image had abnormal behavior, therefore, the target group can be analyzed based on the first attribute data and the second attribute data to determine the target object from the target group.

[0110] It is easy to notice that in the embodiment of the present invention, by using the input images in different time periods as the detection targets, the input image in the first time period is detected, the first attribute data of the target group is obtained from the target area shown in the above input image, and the second attribute data is obtained after detecting the historical image in the second time period. Since the first attribute data and the second attribute data are input images in different time stages respectively, and the second time period corresponding to the second attribute data, therefore, the target group can be analyzed based on the first attribute data and the second attribute data, paying more attention to the interaction effect of the group intelligence evolution behavior of the surrounding people, and thus the accuracy of determining the target object from the target group can be achieved.

[0111] Since the embodiment of the present invention pays more attention to the interaction effect of the group intelligence evolution behavior of the surrounding people, it achieves the purpose of improving the accuracy and detection efficiency of detecting the target object, thus realizing the technical effect of more effectively performing security warning on the target object, and further solving the technical problem that in the pedestrian detection method in the prior art, the analysis of pedestrian features ignores the interaction effect of the group intelligence evolution behavior of the surrounding people.

[0112] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0113] Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a computer-readable storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions for causing a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present invention.

[0114] Embodiment 2

[0115] According to an embodiment of the present invention, there is also provided an apparatus embodiment for implementing the above-mentioned method for processing a target object based on swarm intelligence. Figure 7 As shown in FIG. 7, a schematic structural diagram of a target object processing apparatus based on swarm intelligence according to an embodiment of the present invention, the above-mentioned apparatus includes:

[0116] A detection module 700, configured to detect an input image within a first time period, and obtain first attribute data of a target group from a target area displayed in the above input image, where the first attribute data is used to describe the position of each target individual in the target group in the above input image and the number of individuals included in the target group; a processing module 702, configured to determine a target object from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0117] It should be noted here that the above-mentioned detection module 700 and processing module 702 correspond to steps S202 to S204 in Embodiment 1. The two modules have the same implemented examples and application scenarios as the corresponding steps, but are not limited to the content disclosed in the above-mentioned Embodiment 1. It should be noted that the above-mentioned modules, as part of the apparatus, can run in the computer terminal 10 provided in Embodiment 1.

[0118] It should be noted that the preferred implementation of this embodiment can refer to the relevant description in Embodiment 1, which will not be elaborated here.

[0119] Embodiment 3

[0120] According to an embodiment of the present invention, there is also provided an embodiment of an electronic device, which can be any one of the computing devices in a computing device cluster. The electronic device includes: a processor and a memory, where:

[0121] a processor; and a memory connected to the above-mentioned processor, for providing instructions for the above-mentioned processor to process the following steps: detecting an input image within a first time period, and obtaining first attribute data of a target group from a target area displayed in the above-mentioned input image, where the above-mentioned first attribute data is used to describe the position of each target individual in the above-mentioned target group in the above-mentioned input image and the number of individuals included in the above-mentioned target group; based on the above-mentioned first attribute data and second attribute data, determining a target object from the above-mentioned target group, where the above-mentioned second attribute data is obtained by detecting a historical image within a second time period, and the above-mentioned second time period is the previous time period of the above-mentioned first time period, and the above-mentioned target object has abnormal behavior.

[0122] It is easy to note that in the embodiment of the present invention, by using input images of different time periods as detection targets, detecting the input image within the first time period, obtaining first attribute data of the target group from the target area displayed in the above-mentioned input image, and obtaining second attribute data by detecting the historical image within the second time period, since the above-mentioned first attribute data and second attribute data are input images of different time stages respectively, and in the second time period corresponding to the second attribute data, the target object in the collected historical image has abnormal behavior, therefore, the above-mentioned target group can be analyzed based on the first attribute data and the second attribute data, paying more attention to the interaction effects of the group intelligence evolution behaviors of the surrounding people, and thus the accuracy of determining the target object from the above-mentioned target group can be achieved.

[0123] Since the embodiment of the present invention pays more attention to the interaction effects of the group intelligence evolution behaviors of the surrounding people of the target object, the purpose of improving the accuracy and detection efficiency of detecting the target object is achieved, thereby realizing the technical effect of more effectively performing security warning on the target object, and further solving the technical problem that in the pedestrian detection method in the prior art, the analysis of pedestrian features ignores the interaction effects of the group intelligence evolution behaviors of the surrounding people.

[0124] It should be noted that the preferred implementation of this embodiment can refer to the relevant description in Embodiment 1, which will not be elaborated here.

[0125] Embodiment 4

[0126] According to an embodiment of the present invention, an embodiment of a computer terminal may also be provided. The computer terminal may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the above computer terminal may also be replaced with a terminal device such as a mobile terminal.

[0127] Optionally, in this embodiment, the above computer terminal may be located in at least one of multiple network devices in a computer network.

[0128] In this embodiment, the above computer terminal may execute program code for the following steps in the method for processing a target object based on swarm intelligence: detecting an input image within a first time period, and obtaining first attribute data of a target group from a target area displayed in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group; determining a target object from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0129] Optionally, Figure 8 is a structural block diagram of a computer terminal according to an embodiment of the present invention. As Figure 8 shown, the computer terminal may include: one or more (only one is shown in the figure) processors 802, a memory 804, and a peripheral interface 806.

[0130] Among them, the memory may be used to store software programs and modules, such as program instructions / modules corresponding to the method and device for processing a target object based on swarm intelligence in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, implements the above method for processing a target object based on swarm intelligence. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory may further include a memory remotely disposed relative to the processor, and these remote memories may be connected to the computer terminal through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0131] The processor can call the information and application programs stored in the memory through the transmission device to execute the following steps: detecting the input image within the first time period, and obtaining the first attribute data of the target group from the target area shown in the above input image, where the first attribute data is used to describe the position of each target individual in the target group in the above input image and the number of individuals included in the target group; determining the target object from the target group based on the first attribute data and the second attribute data, where the second attribute data is obtained by detecting the historical image within the second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0132] Optionally, the processor can also execute the program code of the following steps: predicting the flow trajectory of each target individual in the target group based on the second attribute data to obtain the first flow trajectory; determining the second flow trajectory of each target individual in the target group based on the first attribute data; comparing the second flow trajectory with the first flow trajectory to obtain the trajectory abnormality degree; and determining the target object from the target group using the trajectory abnormality degree.

[0133] Optionally, the processor can also execute the program code of the following steps: analyzing the second attribute data using the swarm intelligence evolution model to obtain the first flow trajectory, where the swarm intelligence evolution model is trained using the historical attribute data in the target area, and the swarm intelligence evolution model is used to predict the flow trajectory of each target individual in the target group.

[0134] Optionally, the processor can also execute the program code of the following steps: when the trajectory abnormality degree is greater than the first threshold, determining the target individual corresponding to the trajectory abnormality degree as the target object.

[0135] Optionally, the processor can also execute the program code of the following steps: detecting the input image, and obtaining the feature data of the target group from the target area, where the feature data is used to describe the visual features of each target individual in the target group; when the trajectory abnormality degree is less than or equal to the first threshold, analyzing the first attribute data and the feature data using the target neural network model to obtain the confidence level of each target individual in the target group, where the target neural network model is trained using the associated data set of the target object, and the target neural network model is used to calculate the confidence level of whether each target individual in the target group has abnormal behavior; when the confidence level is greater than the second threshold, determining the target individual corresponding to the confidence level as the target object.

[0136] Optionally, the above-mentioned processor may also execute the program code of the following steps: perform action recognition on the above-mentioned input image, and obtain the confidence level of whether each target individual in the above-mentioned target group has abnormal behavior from the above-mentioned target area; when the above-mentioned confidence level is greater than a third threshold, determine the target individual corresponding to the above-mentioned confidence level as the above-mentioned target object.

[0137] The processor may call the information and application programs stored in the memory through the transmission device to execute the following steps: obtain the input image within the first time period; detect the above-mentioned input image, obtain the first attribute data of the target group from the target area shown in the above-mentioned input image, and based on the above-mentioned first attribute data and the second attribute data, determine the target object from the above-mentioned target group, where the above-mentioned first attribute data is used to describe the position of each target individual in the above-mentioned target group in the above-mentioned input image and the number of individuals included in the above-mentioned target group, the above-mentioned second attribute data is obtained by detecting the historical image within the second time period, the above-mentioned second time period is the previous time period of the above-mentioned first time period, and the above-mentioned target object has abnormal behavior; report the detection result to the server, where the above-mentioned detection result is used to record the above-mentioned target object with abnormal behavior.

[0138] The processor may call the information and application programs stored in the memory through the transmission device to execute the following steps: receive the input image within the first time period from the client; detect the above-mentioned input image, obtain the first attribute data of the target group from the target area shown in the above-mentioned input image, and based on the above-mentioned first attribute data and the second attribute data, determine the target object from the above-mentioned target group, where the above-mentioned first attribute data is used to describe the position of each target individual in the above-mentioned target group in the above-mentioned input image and the number of individuals included in the above-mentioned target group, the above-mentioned second attribute data is obtained by detecting the historical image within the second time period, the above-mentioned second time period is the previous time period of the above-mentioned first time period, and the above-mentioned target object has abnormal behavior; return the detection result to the above-mentioned client and display the above-mentioned detection result on the graphical user interface of the above-mentioned client, where the above-mentioned detection result is used to record the above-mentioned target object with abnormal behavior.

[0139] An embodiment of the present invention provides a solution for processing target objects based on swarm intelligence. By detecting an input image within a first time period, first attribute data of a target group is obtained from a target area shown in the input image, where the first attribute data is used to describe the positions of each target individual in the target group in the input image and the number of individuals included in the target group; based on the first attribute data and second attribute data, a target object is determined from the target group, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period, and the target object has abnormal behavior.

[0140] It is easy to notice that in the embodiment of the present invention, by using input images in different time periods as detection targets, detecting the input image within the first time period, obtaining the first attribute data of the target group from the target area shown in the input image, and obtaining the second attribute data by detecting the historical image within the second time period. Since the first attribute data and the second attribute data are input images in different time stages respectively, and in the second time period corresponding to the second attribute data, the target object in the collected historical image has abnormal behavior, therefore, the target group can be analyzed based on the first attribute data and the second attribute data, paying more attention to the interactive influence of the swarm intelligence evolution behavior of the surrounding people, and further the accuracy of determining the target object from the target group can be achieved.

[0141] Since the embodiment of the present invention pays more attention to the interactive influence of the swarm intelligence evolution behavior of the surrounding people, the purpose of improving the accuracy and detection efficiency of detecting the target object is achieved, thus realizing the technical effect of more effectively performing security warning on the target object, and further solving the technical problem that in the existing pedestrian detection method, the analysis of pedestrian features ignores the interactive influence of the swarm intelligence evolution behavior of the surrounding people.

[0142] Those of ordinary skill in the art can understand that Figure 8 the structure shown is only for illustration, and the computer terminal can also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a palm computer, and terminal devices such as Mobile Internet Devices (MID), PAD, etc. Figure 8 It does not limit the structure of the above electronic device. For example, the computer terminal 8 may further include more or fewer components (such as a network interface, a display device, etc.) than those shown Figure 8 in the figure, or have a different configuration from that shown Figure 8 in the figure.

[0143] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, and the computer-readable storage medium can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, an optical disk, etc.

[0144] Embodiment 5

[0145] According to an embodiment of the present invention, an embodiment of a computer-readable storage medium is also provided. Optionally, in this embodiment, the above computer-readable storage medium can be used to store the program code executed by the method for processing a target object based on swarm intelligence provided in the above Embodiment 1.

[0146] Optionally, in this embodiment, the above computer-readable storage medium can be located in any one of the computer terminals in a computer terminal group in a computer network, or in any one of the mobile terminals in a mobile terminal group.

[0147] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: detecting an input image within a first time period, and obtaining first attribute data of a target group from a target area displayed in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group; determining a target object from the target group based on the first attribute data and second attribute data, where the second attribute data is obtained by detecting a historical image within a second time period, the second time period being the previous time period of the first time period, and the target object has abnormal behavior.

[0148] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: predicting a flow trajectory of each target individual in the target group based on the second attribute data to obtain a first flow trajectory; determining a second flow trajectory of each target individual in the target group based on the first attribute data; comparing the second flow trajectory with the first flow trajectory to obtain a trajectory abnormality degree; and determining the target object from the target group using the trajectory abnormality degree.

[0149] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: analyzing the second attribute data by using a swarm intelligence evolution model to obtain the first flow trajectory, where the swarm intelligence evolution model is trained by using the historical attribute data in the target area, and the swarm intelligence evolution model is used to predict the flow trajectory of each target individual in the target group.

[0150] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: when the trajectory abnormality degree is greater than a first threshold, determining the target individual corresponding to the trajectory abnormality degree as the target object.

[0151] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: detecting the input image, and obtaining the feature data of the target group from the target area, where the feature data is used to describe the visual features of each target individual in the target group; when the trajectory abnormality degree is less than or equal to the first threshold, analyzing the first attribute data and the feature data by using a target neural network model to obtain the confidence degree of each target individual in the target group, where the target neural network model is trained by using the associated data set of the target object, and the target neural network model is used to calculate the confidence degree of whether each target individual in the target group has an abnormal behavior; when the confidence degree is greater than a second threshold, determining the target individual corresponding to the confidence degree as the target object.

[0152] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: performing action recognition on the input image, and obtaining the confidence degree of whether each target individual in the target group has an abnormal behavior; when the confidence degree is greater than a third threshold, determining the target individual corresponding to the confidence degree as the target object.

[0153] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: obtaining an input image within a first period; detecting the input image, obtaining first attribute data of a target group from a target area shown in the input image, and determining a target object from the target group based on the first attribute data and second attribute data, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group, the second attribute data is obtained by detecting historical images within a second period, the second period is the previous period of the first period, and the target object has abnormal behavior; reporting the detection result to a server, where the detection result is used to record the target object with abnormal behavior.

[0154] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: receiving an input image within a first period from a client; detecting the input image, obtaining first attribute data of a target group from a target area shown in the input image, and determining a target object from the target group based on the first attribute data and second attribute data, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group, the second attribute data is obtained by detecting historical images within a second period, the second period is the previous period of the first period, and the target object has abnormal behavior; returning the detection result to the client and displaying the detection result on a graphical user interface of the client, where the detection result is used to record the target object with abnormal behavior.

[0155] The serial numbers of the embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0156] In the above embodiments of the present invention, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0157] In several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the units or modules can be in an electrical or other form.

[0158] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0159] In addition, each functional unit in various embodiments of the present invention may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0160] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned computer-readable storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0161] The above description is only a preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for processing target objects based on swarm intelligence, characterized in that Including: Detecting an input image within a first time period, and obtaining first attribute data and feature data of a target group from a target area shown in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group, and the feature data is used to describe the visual features of each target individual in the target group; Based on the first attribute data and second attribute data, determining the trajectory anomaly degree of each target individual in the target group, where the second attribute data is obtained by detecting a historical image within a second time period, and the second time period is the previous time period of the first time period; In response to the trajectory anomaly degree being less than or equal to a first threshold, determining the confidence degree of each target individual in the target group based on the first attribute data and the feature data; In response to the confidence degree being greater than a second threshold, determining the target individual corresponding to the confidence degree as a target object, where the target object has abnormal behavior.

2. The method for processing a target object based on swarm intelligence according to claim 1, wherein Determining the trajectory anomaly degree of each target individual in the target group based on the first attribute data and the second attribute data includes: Predicting the flow trajectory of each target individual in the target group based on the second attribute data to obtain a first flow trajectory; Determining a second flow trajectory of each target individual in the target group based on the first attribute data; Comparing the second flow trajectory with the first flow trajectory to obtain the trajectory anomaly degree.

3. The method for processing a target object based on swarm intelligence according to claim 2, wherein, Predicting the flow trajectory of each target individual in the target group based on the second attribute data to obtain the first flow trajectory includes: Analyzing the second attribute data by using a swarm intelligence evolution model to obtain the first flow trajectory, where the swarm intelligence evolution model is trained by using historical attribute data in the target area, and the swarm intelligence evolution model is used to predict the flow trajectory of each target individual in the target group.

4. The method for processing a target object based on swarm intelligence according to claim 2, wherein The method for processing a target object based on swarm intelligence further includes: When the trajectory anomaly degree is greater than the first threshold, determining the target individual corresponding to the trajectory anomaly degree as the target object.

5. The method for processing a target object based on swarm intelligence according to claim 2, wherein The method for processing a target object based on swarm intelligence further includes: When the trajectory anomaly degree is less than or equal to the first threshold, analyzing the first attribute data and the feature data by using a target neural network model to obtain the confidence degree of each target individual in the target group, where the target neural network model is trained by using an associated data set of the target object, and the target neural network model is used to calculate the confidence degree of whether each target individual in the target group has abnormal behavior; When the confidence degree is greater than the second threshold, determining the target individual corresponding to the confidence degree as the target object.

6. The method for processing a target object based on swarm intelligence according to claim 5, wherein The second threshold is set based on the trajectory anomaly degree.

7. The method for processing a target object based on swarm intelligence according to claim 1, wherein The method for processing a target object based on swarm intelligence further includes: Performing action recognition on the input image, and obtaining the confidence degree of whether each target individual in the target group has abnormal behavior from the target area. When the confidence level is greater than a third threshold, determine the target individual corresponding to the confidence level as the target object.

8. A method for processing target objects based on swarm intelligence, characterized in that, Including: Obtain an input image within a first time period; Detect the input image, obtain first attribute data and feature data of a target group from a target area shown in the input image, and determine a trajectory anomaly degree of each target individual in the target group based on the first attribute data and second attribute data, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group, the feature data is used to describe the visual features of each target individual in the target group, the second attribute data is obtained by detecting historical images within a second time period, and the second time period is the previous time period of the first time period; in response to the trajectory anomaly degree being less than or equal to a first threshold, determine the confidence level of each target individual in the target group based on the first attribute data and the feature data; in response to the confidence level being greater than a second threshold, determine the target individual corresponding to the confidence level as the target object, where the target object has abnormal behavior; Report the detection result to the server, where the detection result is used to record the target object with abnormal behavior.

9. A method for processing target objects based on swarm intelligence, characterized in that, Including: Receive an input image within a first time period from a client; Detect the input image, obtain first attribute data and feature data of a target group from a target area shown in the input image, and determine a trajectory anomaly degree of each target individual in the target group based on the first attribute data and second attribute data, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group, the feature data is used to describe the visual features of each target individual in the target group, the second attribute data is obtained by detecting historical images within a second time period, and the second time period is the previous time period of the first time period; in response to the trajectory anomaly degree being less than or equal to a first threshold, determine the confidence level of each target individual in the target group based on the first attribute data and the feature data; in response to the confidence level being greater than a second threshold, determine the target individual corresponding to the confidence level as the target object, where the target object has abnormal behavior; Return the detection result to the client and display the detection result on a graphical user interface of the client, where the detection result is used to record the target object with abnormal behavior.

10. An object processing device based on swarm intelligence, characterized in that, Including: A detection module, configured to detect an input image within a first time period, and obtain first attribute data and feature data of a target group from a target area shown in the input image, where the first attribute data is used to describe the position of each target individual in the target group in the input image and the number of individuals included in the target group, and the feature data is used to describe the visual features of each target individual in the target group; A processing module, configured to determine the trajectory anomaly degree of each target individual in the target group based on the first attribute data and the second attribute data, where the second attribute data is obtained by detecting historical images in a second period, and the second period is the period before the first period; in response to the trajectory anomaly degree being less than or equal to a first threshold, determine the confidence level of each target individual in the target group based on the first attribute data and the feature data; in response to the confidence level being greater than a second threshold, determine the target individual corresponding to the confidence level as a target object, where the target object has abnormal behavior.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, where, when the program runs, it controls the device where the computer-readable storage medium is located to execute the method for processing a target object based on swarm intelligence according to any one of claims 1 to 9.

12. An electronic device, characterized in that, Comprising: A processor; And A memory, connected to the processor, for providing instructions for the processor to perform the following processing steps: Detect an input image in a first period, and obtain the first attribute data and the feature data of a target group from a target area displayed in the input image, where the first attribute data is used to describe the position of each target individual in the input image in the target group and the number of individuals included in the target group, and the feature data is used to describe the visual features of each target individual in the target group; Based on the first attribute data and the second attribute data, determine the trajectory anomaly degree of each target individual in the target group, where the second attribute data is obtained by detecting historical images in a second period, and the second period is the period before the first period; In response to the trajectory anomaly degree being less than or equal to a first threshold, determine the confidence level of each target individual in the target group based on the first attribute data and the feature data; In response to the confidence level being greater than a second threshold, determine the target individual corresponding to the confidence level as a target object, where the target object has abnormal behavior.

Images