An ai-based road object detection and tracking system and method

Abstract

The application provides an AI-based road target detection and tracking system and method, wherein the system comprises: a disappearance position acquisition module configured to acquire a disappearance position of a target person; a disguise route information determination module configured to acquire, based on a preset information acquisition template, disguise route information near the disappearance position from a preset city map; a personnel feature acquisition module configured to acquire personnel features of the target person; a disguised personnel feature determination module configured to input the personnel features and the disguise route information into a preset disguised personnel feature AI prediction model to determine disguised personnel features; a first road image acquisition module configured to acquire a first road image near the disappearance position; and a target person detection module configured to detect the target person from the first road image based on the personnel features and the disguised personnel features. The application can avoid the problem that the target person may disguise himself by using the disguise route of the surrounding environment and thus be lost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a road target detection and tracking system and method based on AI. BACKGROUND

[0002] At present, when the responsible person performs the tracking task of the target person, the target person may use the camouflage way of the surrounding environment to camouflage himself, so that the responsible person cannot identify the camouflaged target person, thereby losing the target person. For example, the responsible person tracks the target person to a shopping mall, the target person enters a clothing store in the shopping mall, changes clothes, wears a hat and other accessories to complete the camouflage, and then walks out of the clothing store. However, the responsible person cannot identify the camouflaged target person at the moment and loses the target person on the spot. In addition, there is also a method of tracking the target person through monitoring facilities. Similarly, when the target person is camouflaged, the tracking through the monitoring facilities will also lose the target person.

[0003] Therefore, a solution is urgently needed. SUMMARY

[0004] One of the purposes of the present application is to provide a road target detection and tracking system based on AI, which can avoid the problem of losing the target person caused by the target person using the camouflage way of the surrounding environment to camouflage himself.

[0005] The road target detection and tracking system based on AI provided by the embodiment of the present application comprises:

[0006] A disappearance position acquisition module is configured to acquire the disappearance position of the target person.

[0007] A camouflage way information determination module is configured to acquire the camouflage way information near the disappearance position from a preset city map based on a preset information acquisition template.

[0008] A personnel feature acquisition module is configured to acquire the personnel feature of the target person.

[0009] A camouflaged personnel feature determination module is configured to input the personnel feature and the camouflage way information into a preset camouflaged personnel feature AI prediction model to determine the camouflaged personnel feature.

[0010] A first road image acquisition module is configured to acquire the first road image near the disappearance position.

[0011] A target person detection module is configured to detect the target person from the first road image based on the personnel feature and the camouflaged personnel feature.

[0012] A second road image acquisition module is configured to acquire the second road image of the target person when the target person is detected.

[0013] Preferably, the disappearance position obtaining module obtains the disappearance position of the target person, comprising:

[0014] When receiving the lost instruction input by the responsible person based on the smart mobile terminal carried by the responsible person, the terminal position of the smart mobile terminal is obtained and used as the disappearance position;

[0015] And / or,

[0016] When the image tracking of the target person fails, the last appearance position of the target person when the image tracking of the target person is performed is obtained and used as the disappearance position.

[0017] Preferably, the target person detecting module detects the target person from the first road image based on the personnel feature and the disguised personnel feature, comprising:

[0018] Based on the preset feature extraction template, the appearance personnel feature of the appearing personnel in the first road image is extracted;

[0019] The appearance personnel feature is matched with the personnel feature and the disguised personnel feature respectively, and the matching condition is obtained;

[0020] Based on the matching condition and the preset target person confirmation index, the target person confirmation degree of the appearing personnel is determined;

[0021] When the target person confirmation degree is greater than or equal to the preset target person confirmation degree threshold, the corresponding appearing personnel is taken as the target person.

[0022] Preferably, the second road image obtaining module tracks and obtains the second road image of the target person, comprising:

[0023] The personnel position, the moving direction and the moving speed of the target person are obtained;

[0024] Based on the personnel position, the moving direction and the moving speed, a plurality of possible moving routes of the target person moving from the personnel position in the future preset time are predicted in the city map;

[0025] From the city map, a first branch point close to the personnel position and a post-branch route after the branch point formed by each possible moving route are determined;

[0026] From the city map, the face moving position sequence and the face moving direction sequence in the process of walking a preset distance along the post-branch route by the target person are predicted;

[0027] From the city map, the facility position and the monitoring range of the monitoring facility responsible for monitoring the post-branch route are determined;

[0028] Based on the facility position, the monitoring range, the face moving position sequence and the face moving direction sequence, the first shooting angle of the monitoring facility when the target person walks along the post-branch route is determined;

[0029] The monitoring facility is adjusted in advance to the first shooting angle;

[0030] When the personnel position is about to reach the branch point, based on the current personnel position, the moving direction and the moving speed, it is determined that the target person is about to move along the branched route;

[0031] The second road image of the target person is acquired by the monitoring facility responsible for monitoring the branched route along which the target person is about to move.

[0032] Preferably, the second road image acquisition module determines the first shooting angle of the monitoring facility when the target person walks along the branched route based on the facility position, the monitoring range, the sequence of face moving positions and the sequence of face moving directions, comprising:

[0033] The size of the shooting range of the monitoring facility is acquired;

[0034] The target shooting range of the size of the shooting range that meets the shooting range condition is determined from the monitoring range;

[0035] The angle at which the monitoring facility shoots the target shooting range is taken as the first shooting angle;

[0036] The shooting range condition comprises:

[0037] All the face moving positions in the sequence of face moving positions fall within the target shooting range;

[0038] The minimum distance between the frontmost face moving position in the sequence of face moving positions and the boundary of the target shooting range is less than or equal to a preset first distance threshold;

[0039] The minimum distance between the last face moving position in the sequence of face moving positions and the boundary of the target shooting range is greater than or equal to a preset second distance threshold;

[0040] The straight line distance between the middle one or two face moving positions in the sequence of face moving positions and the boundary center of the target shooting range is less than or equal to a preset third distance threshold;

[0041] The direction included angle between the straight line direction from the boundary center of the target shooting range to the facility position and any face moving direction in the sequence of face moving directions falls within a preset included angle interval.

[0042] The AI-based road target detection and tracking method provided by the embodiment of the application comprises:

[0043] Step S1: acquiring the disappearance position of the target person;

[0044] Step S2: obtaining camouflage route information near the disappearance position from the preset city map based on a preset information acquisition template;

[0045] Step S3: obtaining personnel features of the target person;

[0046] Step S4: inputting the personnel features and the camouflage route information into a preset camouflage personnel feature AI prediction model to determine camouflage personnel features;

[0047] Step S5: obtaining a first road image near the disappearance position;

[0048] Step S6: detecting the target person from the first road image based on the personnel features and the camouflage personnel features;

[0049] Step S7: when the target person is detected, tracking a second road image of the target person.

[0050] Preferably, step S1: obtaining the disappearance position of the target person, comprises:

[0051] When receiving a losing track instruction input by the responsible person based on the smart mobile terminal carried by the responsible person, obtaining a terminal position of the smart mobile terminal as the disappearance position;

[0052] and / or,

[0053] When image tracking of the target person fails, obtaining a last appearance position of the target person when the target person is tracked by image as the disappearance position.

[0054] Preferably, step S6: detecting the target person from the first road image based on the personnel features and the camouflage personnel features, comprises:

[0055] Based on a preset feature extraction template, extracting appearance personnel features of the appearing personnel in the first road image;

[0056] Matching the appearance personnel features with the personnel features and the camouflage personnel features respectively to obtain matching conditions;

[0057] Based on the matching conditions and a preset target person confirmation index, determining a target person confirmation degree of the appearing personnel;

[0058] When the target person confirmation degree is greater than or equal to a preset target person confirmation threshold, the corresponding appearing personnel is regarded as the target person.

[0059] Preferably, in step S7, the second road image of the target person is tracked, comprising:

[0060] Obtaining a personnel position, a moving direction and a moving speed of the target person;

[0061] predicting, based on the personnel position, the moving direction and the moving speed, a plurality of possible moving routes of the target person from the personnel position in the city map within a preset future time;

[0062] determining, from the city map, a first branching point close to the personnel position formed by the plurality of possible moving routes and a post-branching route after the branching point;

[0063] predicting, from the city map, a sequence of facial moving positions and a sequence of facial moving directions of the target person in a process of walking a preset distance along the post-branching route;

[0064] determining, from the city map, a facility position and a monitoring range of a monitoring facility responsible for monitoring the post-branching route;

[0065] determining, based on the facility position, the monitoring range, the sequence of facial moving positions and the sequence of facial moving directions, a first shooting angle of the monitoring facility when the target person walks along the post-branching route;

[0066] adjusting the monitoring facility to the first shooting angle in advance;

[0067] when the personnel position is about to reach the branching point, determining that the target person is about to move along the post-branching route based on the current personnel position, the moving direction and the moving speed;

[0068] acquiring a second road image of the target person by the monitoring facility responsible for monitoring the post-branching route along which the target person is about to move.

[0069] Preferably, determining, based on the facility position, the monitoring range, the sequence of facial moving positions and the sequence of facial moving directions, the first shooting angle of the monitoring facility when the target person walks along the post-branching route, comprises:

[0070] acquiring a shooting range size of the monitoring facility;

[0071] determining, from the monitoring range, a target shooting range of the shooting range size satisfying a shooting range condition;

[0072] taking an angle at which the monitoring facility shoots the target shooting range as the first shooting angle;

[0073] wherein the shooting range condition comprises:

[0074] all the facial moving positions in the sequence of facial moving positions fall within the target shooting range;

[0075] a minimum distance between a frontmost facial moving position in the sequence of facial moving positions and a boundary of the target shooting range is less than or equal to a preset first distance threshold;

[0076] A minimum distance between a last face movement position in the sequence of face movement positions and a boundary of the target shooting range is greater than or equal to a preset second distance threshold;

[0077] A straight line distance between a middle one or two face movement positions in the sequence of face movement positions and a boundary center of the target shooting range is less than or equal to a preset third distance threshold;

[0078] A direction included angle between a straight line direction from the boundary center of the target shooting range to the facility position and any one face movement direction in the sequence of face movement directions falls within a preset included angle interval.

[0079] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art upon examination of the following or can be learned by practice of the application. The objects and other advantages of the application can be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

[0080] The technical solutions of the present application will be further described in detail below with the help of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0081] The accompanying drawings are included to provide a further understanding of the present application and are incorporated in and constitute a part of this specification, illustrate embodiments of the present application and serve to explain the present application, and do not limit the present application. In the drawings:

[0082] Figure 1 It is a schematic diagram of an AI-based road target detection and tracking system in an embodiment of the present application;

[0083] Figure 2 It is an application schematic diagram of the second road image of the target person tracked and acquired in an embodiment of the present application;

[0084] Figure 3 It is a schematic diagram of an AI-based road target detection and tracking method in an embodiment of the present application. DETAILED DESCRIPTION

[0085] The preferred embodiments of the present application will be described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to explain and illustrate the present application, and do not limit the present application.

[0086] The present application provides an AI-based road target detection and tracking system, as shown in Figure 1 , comprising:

[0087] The disappearing position acquisition module 1 is configured to acquire the disappearing position of the target person.

[0088] The camouflage route information determination module 2 is configured to acquire camouflage route information near the disappearance position from a preset city map based on a preset information acquisition template; the camouflage route information includes facilities and places that can be used for the target person to camouflage, such as clothing stores (for changing clothes) and parking lots (for driving a car);

[0089] The personnel feature acquisition module 3 is configured to acquire personnel features of the target person; the personnel features include gender, height, body shape, skin color, walking habit posture, and clothes on the day;

[0090] The camouflage personnel feature determination module 4 is configured to input the personnel features and the camouflage route information into a preset camouflage personnel feature AI prediction model to determine camouflage personnel features; the camouflage personnel feature AI prediction model is an artificial intelligence model obtained by training a neural network model based on a large number of historical records of the target person performing camouflage (such as comparison pictures before and after changing clothes) until the model converges; the model can predict the camouflage personnel features of the target person after the target person performs camouflage based on the personnel features and the camouflage route information; for example, the personnel features include gender, height, body shape, and skin color, and the camouflage route information is a clothing store; the model can output the body shape of the target person after the target person changes clothes (for example, wears a hat and carries an umbrella) in the clothing store;

[0091] The first road image acquisition module 5 is configured to acquire a first road image near the disappearance position;

[0092] The target person detection module 6 is configured to detect the target person from the first road image based on the personnel features and the camouflage personnel features;

[0093] The second road image acquisition module 7 is configured to acquire a second road image of the target person when the target person is detected.

[0094] The working principle and beneficial effects of the above technical solution are as follows:

[0095] The application introduces the camouflage personnel feature AI prediction model, predicts the output camouflage personnel features, detects the target person from the first road image based on the personnel features and the camouflage personnel features, acquires the second road image of the target person when the target person is detected, and realizes target detection and tracking. The problem that the target person may use the camouflage route of the surrounding environment to camouflage himself / herself and be lost can be avoided.

[0096] In one embodiment, the disappearance position acquisition module 1 acquires the disappearance position of the target person, including:

[0097] When receiving a lost instruction input by the responsible person based on the smart mobile terminal carried by the responsible person, the terminal position of the smart mobile terminal is acquired and used as the disappearance position;

[0098] and / or,

[0099] When the image tracking of the target person fails, the last appearance position of the target person when the image tracking of the target person is acquired is acquired as the disappearance position.

[0100] The working principle and beneficial effects of the above technical solution are:

[0101] There are two ways to acquire the disappearance position of the target person: one is that when the person in charge of tracking the target person finds that he / she has lost the target person, the person in charge can input a lost instruction based on the smart mobile terminal carried by the person in charge, and the terminal position of the smart mobile terminal at present is taken as the disappearance position. After the person in charge determines that the target person is lost, the person in charge can one-key track the target person. The other is that when the monitoring facility fails to acquire the image tracking of the target person, the last appearance position of the target person when the image tracking of the target person is acquired is taken as the disappearance position.

[0102] In one embodiment, the target person detection module 6 detects the target person from the first road image based on the personnel feature and the disguised personnel feature, including:

[0103] Based on the preset feature extraction template, the appearance personnel feature of the appearing personnel in the first road image is extracted; the appearance personnel feature includes the gender, height, human contour, skin color, walking habit posture, and clothing of the appearing personnel on the day, etc.

[0104] The appearance personnel feature is matched with the personnel feature and the disguised personnel feature respectively, and the matching condition is acquired; the matching condition is which type of appearance personnel feature is matched with the personnel feature / disguised personnel feature, for example, the height of the appearing personnel is matched with the height of the target person, etc.

[0105] Based on the matching condition and the preset target person confirmation index, the target person confirmation degree of the appearing personnel is determined; in the target person confirmation index, the confirmation degree corresponding to different matching conditions is included. Generally, the more types of appearance personnel features are matched with the personnel feature / disguised personnel feature, the higher the confirmation degree is.

[0106] When the target person confirmation degree is greater than or equal to the preset target person confirmation degree threshold, the corresponding appearing personnel is taken as the target person.

[0107] The working principle and beneficial effects of the above technical solution are:

[0108] When the target person is detected from the first road image, the matching condition and the target person confirmation index are introduced to quickly determine the target person confirmation degree, and the target person is detected from the appearing personnel based on the target person confirmation degree, thereby improving the comprehensiveness and detection efficiency of the target person detection.

[0109] In one embodiment, the second road image acquisition module 7 tracks and acquires the second road image of the target person, as shown in Figure 2 The second road image acquisition module 7 includes:

[0110] acquire the personnel position, the moving direction and the moving speed of the target person; the three can be determined based on the image of the target person captured previously;

[0111] predict a plurality of possible moving routes of the target person moving from the personnel position in a future preset time based on the personnel position, the moving direction and the moving speed; when predicting, determine the starting point of the possible moving route based on the personnel position, determine the direction in which the possible moving route can be extended from the starting point based on the moving direction, and determine the length of the possible moving route based on the moving speed;

[0112] determine the first branch point A close to the personnel position and the post-branch route B after the branch point formed by each possible moving route from the city map;

[0113] predict the face moving position sequence and the face moving direction sequence of the target person in the process of walking a preset distance along the post-branch route from the city map;

[0114] determine the facility position and the monitoring range of the monitoring facility C responsible for monitoring the post-branch route from the city map;

[0115] determine the first shooting angle of the monitoring facility when the target person walks along the post-branch route based on the facility position, the monitoring range, the face moving position sequence and the face moving direction sequence;

[0116] adjust the monitoring facility C to the first shooting angle in advance;

[0117] when the personnel position is about to reach the branch point, determine that the target person is about to move along the post-branch route based on the current personnel position, the moving direction and the moving speed; when determining, take the post-branch route close to the personnel position, pointed by the moving direction, and not reduced by the moving speed as the post-branch route along which the target person is about to move;

[0118] acquire the second road image of the target person by the monitoring facility C responsible for monitoring the post-branch route along which the target person is about to move.

[0119] The working principle and beneficial effects of the above technical solution are:

[0120] When actually tracking, a situation may occur: for example, Figure 2As shown, the target person can move along two bifurcated back routes B next, one bifurcated line B is into alley D, E is an internal passage, if the monitoring facility C does not adjust to the suitable angle of capturing the picture of the target person entering the alley B in advance, the target person can hide from the tracking by the internal passage E as soon as entering the alley D, however, there is an internal passage F in the alley D, therefore, without knowing, it is unknown that the target person hides from the tracking by the internal passage E or the internal passage F, which increases the difficulty of subsequent tracking. The monitoring facility in reality is connected with the wall through a rotating table, and the shooting angle is adjustable. The embodiment of the present application adjusts the monitoring facility C to the first shooting angle in advance, which can ensure that the target person can be tracked immediately as soon as entering the road area monitored by the monitoring facility, avoids that the target person hides from the tracking due to no timely tracking caused by the complex road condition, and is particularly suitable, and is more intelligent.

[0121] In one embodiment, the second road image acquisition module 7 determines the first shooting angle of the monitoring facility when the target person walks along the bifurcated back route based on the facility position, the monitoring range, the face moving position sequence and the face moving direction sequence, including:

[0122] Acquiring the shooting range size of the monitoring facility; the monitoring range is the whole range that can be monitored by the monitoring facility cooperating with the rotating table; the shooting range size is the size of the range that can be shot by the monitoring facility when it is fixed and does not move;

[0123] Determining the target shooting range with the shooting range size meeting the shooting range condition from the monitoring range;

[0124] Taking the angle of the monitoring facility shooting the target shooting range as the first shooting angle;

[0125] Wherein, the shooting range condition includes:

[0126] All the face moving positions in the face moving position sequence fall within the target shooting range; when this condition is met, it can be ensured that the monitoring facility does not need to rotate the rotating table in the early stage after adjusting to the first shooting angle, and can capture all the face images of the target person entering the monitoring range responsible for by the monitoring facility;

[0127] The minimum distance between the first face moving position in the face moving position sequence and the boundary of the target shooting range is less than or equal to the preset first distance threshold; the minimum distance between the last face moving position in the face moving position sequence and the boundary of the target shooting range is greater than or equal to the preset second distance threshold; when the two conditions are met, it can be ensured that more shooting range is left for the picture capture of the target person in the subsequent under the premise of capturing all the face images of the target person entering the monitoring range responsible for by the monitoring facility, which improves the utilization efficiency of shooting resources;

[0128] The straight line distance between the middle one or two face moving positions in the face moving position sequence and the boundary center of the target shooting range is less than or equal to a preset third distance threshold; when this condition is met, it can be ensured that the target person moves up, down, left and right after entering the monitoring range responsible for by the monitoring facility, and the monitoring facility can track as much as possible without rotating the turntable;

[0129] The direction angle between the straight line direction from the boundary center of the target shooting range to the facility position and any one face moving direction in the face moving direction sequence falls within a preset angle interval; when this condition is met, it can be ensured that the monitoring facility can shoot as much as possible the front face image of the target person when the target person enters the monitoring range responsible for by the monitoring facility. The angle interval can be 0 degrees to 75 degrees.

[0130] The working principle and beneficial effects of the above technical solutions are as follows:

[0131] Based on the facility position, the monitoring range, the face moving position sequence and the face moving direction sequence, the first shooting angle of the monitoring facility when the target person walks along the bifurcated route is determined, the shooting range condition is introduced, and the comprehensiveness and suitability of the first shooting angle determination are improved.

[0132] The embodiment of the application provides a road target detection and tracking method based on AI, as shown in Figure 3 The method comprises the following steps:

[0133] Step S1: obtaining the disappearance position of the target person;

[0134] Step S2: obtaining the disguised route information near the disappearance position from the preset city map based on a preset information acquisition template;

[0135] Step S3: obtaining the personnel characteristics of the target person;

[0136] Step S4: inputting the personnel characteristics and the disguised route information into a preset disguised personnel characteristic AI prediction model to determine the disguised personnel characteristics;

[0137] Step S5: obtaining a first road image near the disappearance position;

[0138] Step S6: detecting the target person from the first road image based on the personnel characteristics and the disguised personnel characteristics;

[0139] Step S7: when the target person is detected, a second road image of the target person is tracked and obtained.

[0140] In one embodiment, step S1: obtaining the disappearance position of the target person, comprises:

[0141] When receiving the lost instruction input by the responsible person based on the smart mobile terminal carried by the person, the terminal position of the smart mobile terminal is acquired and taken as the disappearance position;

[0142] and / or,

[0143] When the image tracking acquisition on the target person fails, the last appearance position of the target person when the image tracking is performed on the target person is acquired and taken as the disappearance position.

[0144] In one embodiment, the step S6 of detecting the target person from the first road image based on the personnel feature and the disguised personnel feature comprises:

[0145] Based on the preset feature extraction template, the appearance personnel feature of the appearance personnel in the first road image is extracted;

[0146] The appearance personnel feature is matched with the personnel feature and the disguised personnel feature respectively, and the matching condition is acquired;

[0147] Based on the matching condition and the preset target person confirmation index, the target person confirmation degree of the appearance personnel is determined;

[0148] When the target person confirmation degree is greater than or equal to the preset target person confirmation degree threshold, the corresponding appearance personnel is taken as the target person.

[0149] In one embodiment, in the step S7, the second road image of the target person is tracked and acquired, which comprises:

[0150] The personnel position, the moving direction and the moving speed of the target person are acquired;

[0151] Based on the personnel position, the moving direction and the moving speed, a plurality of possible moving routes of the target person moving from the personnel position in a future preset time are predicted in the city map;

[0152] From the city map, a first branch point close to the personnel position and a post-branch route after the branch point formed by each possible moving route are determined;

[0153] From the city map, a face moving position sequence and a face moving direction sequence in the process of the target person walking a preset distance along the post-branch route are predicted;

[0154] From the city map, the facility position and the monitoring range of a monitoring facility responsible for monitoring the post-branch route are determined;

[0155] Based on the facility position, the monitoring range, the face moving position sequence and the face moving direction sequence, a first shooting angle of the monitoring facility when the target person walks along the post-branch route is determined;

[0156] The monitoring facility is adjusted to the first shooting angle in advance;

[0157] When the personnel position is about to reach the branch point, based on the current personnel position, the moving direction and the moving speed, it is determined that the target person is about to move along the branch rear route;

[0158] The second road image of the target person is tracked by the monitoring facility responsible for monitoring the branch rear route along which the target person is about to move.

[0159] In an embodiment, based on the facility position, the monitoring range, the sequence of face moving positions and the sequence of face moving directions, the first shooting angle of the monitoring facility when the target person walks along the branch rear route is determined, comprising:

[0160] The size of the shooting range of the monitoring facility is obtained;

[0161] From the monitoring range, a target shooting range with a size of the shooting range satisfying a shooting range condition is determined;

[0162] The angle at which the monitoring facility shoots the target shooting range is taken as the first shooting angle;

[0163] The shooting range condition comprises:

[0164] All face moving positions in the sequence of face moving positions fall within the target shooting range;

[0165] The minimum distance between the frontmost face moving position in the sequence of face moving positions and the boundary of the target shooting range is less than or equal to a preset first distance threshold;

[0166] The minimum distance between the last face moving position in the sequence of face moving positions and the boundary of the target shooting range is greater than or equal to a preset second distance threshold;

[0167] The straight line distance between the middle one or two face moving positions in the sequence of face moving positions and the boundary center of the target shooting range is less than or equal to a preset third distance threshold;

[0168] The direction included angle between the straight line direction from the boundary center of the target shooting range to the facility position and any face moving direction in the sequence of face moving directions falls within a preset included angle interval.

[0169] It needs to be particularly pointed out that the target person data (such as personnel features, road images, etc.) obtained and processed by the embodiments of the present application when performing road target detection and tracking conforms to the relevant laws and regulations of the state on data security and personal information protection. Before actually deploying and implementing the system or method, the authorization or consent of the relevant users will be obtained in accordance with the law.

[0170] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. An AI-based road target detection and tracking system, characterized in that, include: The "Disappearance Location Acquisition Module" is used to obtain the disappearance location of the target person. The camouflage route information determination module is used to obtain camouflage route information near the disappearance location from a preset city map based on a preset information acquisition template; The personnel feature acquisition module is used to acquire the personnel features of the target person; The disguised personnel feature determination module is used to input the personnel features and the disguise route information into a preset disguised personnel feature AI prediction model to determine the disguised personnel features; The first road image acquisition module is used to acquire a first road image near the disappearance location; A target person detection module is used to detect the target person from the first road image based on the person's characteristics and the disguised person's characteristics; The second road image acquisition module is used to track and acquire a second road image of the target person when the target person is detected.

2. The AI-based road target detection and tracking system as described in claim 1, characterized in that, The disappearance location acquisition module acquires the disappearance location of the target person, including: When a tracking command is received from the person in charge based on a smart mobile terminal they are carrying, the location of the smart mobile terminal is obtained and used as the location of disappearance. And / or, When image tracking of the target person fails, the last appearance position of the target person during image tracking is obtained and used as the disappearance position.

3. The AI-based road target detection and tracking system as described in claim 1, characterized in that, The target person detection module detects the target person from the first road image based on the person's characteristics and the disguised person's characteristics, including: Based on a preset feature extraction template, extract the features of people appearing in the first road image; The characteristics of the appearing personnel are matched with the characteristics of the actual personnel and the characteristics of the disguised personnel, respectively, to obtain the matching results; Based on the matching results and the preset target person confirmation index, the target person confirmation degree of the appearing person is determined; When the target person confirmation rate is greater than or equal to the preset target person confirmation rate threshold, the corresponding person who appears will be regarded as the target person.

4. The AI-based road target detection and tracking system as described in claim 1, characterized in that, The second road image acquisition module tracks and acquires a second road image of the target person, including: Obtain the target person's location, direction of movement, and speed of movement; Based on the person's location, direction of movement, and speed of movement, multiple possible movement routes of the target person from the person's location within a preset time period are predicted in the city map. From the city map, determine the first fork point near the location of the person formed by each possible movement route, and the subsequent forks after the fork point; Predict from the city map the sequence of facial movement positions and the sequence of facial movement directions as the target person walks a preset distance along the forked route; Determine the location and monitoring range of the monitoring facilities responsible for monitoring the branching route from the city map; Based on the facility location, monitoring range, facial movement position sequence, and facial movement direction sequence, determine the first shooting angle of the monitoring facility when the target person walks along the forked route; The monitoring equipment is pre-adjusted to the initial shooting angle; When the person's location is about to reach the fork point, based on the current person's location, direction of movement, and speed of movement, it is determined that the target person will follow the route after the fork in the movement; A second road image of the target person is obtained by tracking the target person along the moving fork in the road, as monitored by the surveillance facility.

5. The AI-based road target detection and tracking system as described in claim 4, characterized in that, The second road image acquisition module determines the initial shooting angle of the monitoring facility when the target person walks along the forked route, based on the facility location, monitoring range, facial movement position sequence, and facial movement direction sequence, including: Obtain the size of the shooting range of the monitoring facility; Determine a target shooting range from the monitoring range that meets the shooting range conditions in terms of shooting range size; The angle at which the monitoring facility captures the target's shooting range is taken as the initial shooting angle; The shooting range conditions include: All facial movement positions in the facial movement position sequence fall within the target shooting range; The minimum distance between the foremost facial movement position in the facial movement position sequence and the boundary of the target shooting range is less than or equal to a preset first distance threshold. The minimum distance between the last face movement position in the face movement position sequence and the boundary of the target shooting range is greater than or equal to a preset second distance threshold. The straight-line distance between the middle one or two face movement positions in the face movement position sequence and the boundary center of the target shooting range is less than or equal to a preset third distance threshold. The angle between the straight line from the center of the target shooting range boundary to the facility location and any face movement direction in the face movement direction sequence is within a preset angle range.

6. A road target detection and tracking method based on AI, characterized in that, include: Step S1: Obtain the location where the target person disappeared; Step S2: Based on a preset information acquisition template, obtain the disguised route information near the disappearance location from a preset city map; Step S3: Obtain the personnel characteristics of the target person; Step S4: Input the personnel characteristics and the disguise route information into the preset disguised personnel characteristic AI prediction model to determine the disguised personnel characteristics; Step S5: Obtain a first road image near the disappearance location; Step S6: Detect the target person from the first road image based on the person's characteristics and the disguised person's characteristics; Step S7: When detected, track and acquire a second road image of the target person.

7. The AI-based road target detection and tracking method as described in claim 6, characterized in that, Step S1: Obtaining the location where the target person disappeared includes: When a tracking command is received from the person in charge based on a smart mobile terminal they are carrying, the location of the smart mobile terminal is obtained and used as the location of disappearance. And / or, When image tracking of the target person fails, the last appearance position of the target person during image tracking is obtained and used as the disappearance position.

8. The road target detection and tracking method based on AI as described in claim 6, characterized in that, Step S6: Detecting the target person from the first road image based on the person's characteristics and the disguised person's characteristics, including: Based on a preset feature extraction template, extract the features of people appearing in the first road image; The characteristics of the appearing personnel are matched with the characteristics of the actual personnel and the characteristics of the disguised personnel, respectively, to obtain the matching results; Based on the matching results and the preset target person confirmation index, the target person confirmation degree of the appearing person is determined; When the target person confirmation rate is greater than or equal to the preset target person confirmation rate threshold, the corresponding person who appears will be regarded as the target person.

9. The AI-based road target detection and tracking method as described in claim 6, characterized in that, In step S7, tracking and acquiring the second road image of the target person includes: Obtain the target person's location, direction of movement, and speed of movement; Based on the person's location, direction of movement, and speed of movement, multiple possible movement routes of the target person from the person's location within a preset time period are predicted in the city map. From the city map, determine the first fork point near the location of the person formed by each possible movement route, and the subsequent forks after the fork point; Predict from the city map the sequence of facial movement positions and the sequence of facial movement directions as the target person walks a preset distance along the forked route; Determine the location and monitoring range of the monitoring facilities responsible for monitoring the branching route from the city map; Based on the facility location, monitoring range, facial movement position sequence, and facial movement direction sequence, determine the first shooting angle of the monitoring facility when the target person walks along the forked route; The monitoring equipment is pre-adjusted to the initial shooting angle; When the person's location is about to reach the fork point, based on the current person's location, direction of movement, and speed of movement, it is determined that the target person will follow the route after the fork in the movement; A second road image of the target person is obtained by tracking the target person along the moving fork in the road, as monitored by the surveillance facility.

10. The AI-based road target detection and tracking method as described in claim 9, characterized in that, The determination of the initial shooting angle of the monitoring facility when the target person walks along the forked route, based on the facility location, monitoring range, facial movement position sequence, and facial movement direction sequence, includes: Obtain the size of the shooting range of the monitoring facility; Determine a target shooting range from the monitoring range that meets the shooting range conditions in terms of shooting range size; The angle at which the monitoring facility captures the target's shooting range is taken as the initial shooting angle; The shooting range conditions include: All facial movement positions in the facial movement position sequence fall within the target shooting range; The minimum distance between the foremost facial movement position in the facial movement position sequence and the boundary of the target shooting range is less than or equal to a preset first distance threshold. The minimum distance between the last face movement position in the face movement position sequence and the boundary of the target shooting range is greater than or equal to a preset second distance threshold. The straight-line distance between the middle one or two face movement positions in the face movement position sequence and the boundary center of the target shooting range is less than or equal to a preset third distance threshold. The angle between the straight line from the center of the target shooting range boundary to the facility location and any face movement direction in the face movement direction sequence is within a preset angle range.

Images