Target vehicle identification method and device, electronic equipment and readable storage medium
By obtaining the vehicle's picture and trajectory data and combining similarity screening, the problem of low vehicle recognition accuracy is solved and the accuracy of vehicle recognition is improved.
Patent Information
- Application Number
- CN202410104535.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-25
- Publication Date
- 2025-07-25
AI Technical Summary
In the prior art, the vehicle identification accuracy rate is low, and it cannot be effectively identified especially when the license plate is blocked or unclear.
By obtaining the image data and trajectory data of the target vehicle, combining the similarity between the detection trajectory and trajectory data of the vehicle in the detection data set and the similarity between the detection pictures and the image data, vehicles with similarity matching the threshold are selected and their vehicle identification is marked to improve the recognition accuracy.
Through the combined screening of detection trajectory and picture data, the accuracy of vehicle identification is improved and the vehicle identification is ensured closer to the actual driving situation.
Smart Images
Figure CN120375299A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of transportation technologies, and in particular, to a method, apparatus, electronic device, and readable storage medium for identifying a target vehicle. Background Art
[0002] The identification of a vehicle's identity is achieved based on the vehicle license plate, which is an important part of traffic management. In the related art, the identification of a vehicle's identity usually involves a camera capturing the vehicle's license plate and then identifying the license plate to confirm the vehicle's license plate. However, in the related art, there are situations where the vehicle's license plate is blocked or the captured license plate is unclear. In such cases, the captured license plate cannot be effectively identified, resulting in a low accuracy rate for vehicle identification.
[0003] It can be seen that there is a problem of low accuracy rate for vehicle identification in the related art. Summary of the Invention
[0004] Embodiments of this application provide a method, apparatus, electronic device, and readable storage medium for identifying a target vehicle to solve the problem of low accuracy rate for vehicle identification in the related art.
[0005] To solve the above problems, this application is implemented as follows:
[0006] In a first aspect, an embodiment of this application provides a method for identifying a target vehicle, including:
[0007] Obtain the picture data and trajectory data of a target vehicle, as well as a first detection data set, where the first detection data set includes the vehicle identifier, detection picture, and detection trajectory corresponding to each vehicle among N vehicles within a first set time, and N is a positive integer greater than 1;
[0008] Identify the vehicles in the first detection data set whose corresponding detection trajectories satisfy a first threshold with the trajectory data to obtain a first intermediate set;
[0009] Identify the vehicles in the first intermediate set whose similarity between the corresponding detection pictures and the picture data is greater than a second similarity threshold to obtain a first set;
[0010] Mark the vehicle identifier of a vehicle in the first set as the vehicle identifier of the target vehicle.
[0011] In a second aspect, an embodiment of this application further provides a device for identifying a target vehicle, including:
[0012] A first acquisition module, configured to acquire image data, trajectory data of a target vehicle, and a first detection data set, where the first detection data set includes vehicle identifiers, detection images, and detection trajectories corresponding to each of N vehicles within a first set time, and N is a positive integer greater than 1;
[0013] A first recognition module, configured to recognize vehicles in the first detection data set whose corresponding detection trajectories satisfy a first threshold with the trajectory data, to obtain a first intermediate set;
[0014] A second recognition module, configured to recognize vehicles in the first intermediate set whose similarity between the corresponding detection images and the image data is greater than a second similarity threshold, to obtain a first set;
[0015] A marking module, configured to mark the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle.
[0016] In a third aspect, an embodiment of the present application further provides an electronic device, including a processor, a memory, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps in the target vehicle recognition method described in the first aspect above.
[0017] In a fourth aspect, an embodiment of the present application further provides a readable storage medium for storing a program, which when executed by a processor implements the steps in the target vehicle recognition method described in the first aspect above.
[0018] In the embodiment of the present application, by acquiring image data, trajectory data of a target vehicle, and a first detection data set, where the first detection data set includes vehicle identifiers, detection images, and detection trajectories corresponding to each of N vehicles within a first set time, and N is a positive integer greater than 1; recognizing vehicles in the first detection data set whose corresponding detection trajectories satisfy a first threshold with the trajectory data, to obtain a first intermediate set; recognizing vehicles in the first intermediate set whose similarity between the corresponding detection images and the image data is greater than a second similarity threshold, to obtain a first set; marking the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle. In this way, by combining the detection trajectory and the trajectory data, and the detection image and the image data, the first set is screened from the first detection data set, so as to mark the target vehicle through the vehicle identifier of the vehicle in the first set, making the vehicle identifier of the target vehicle closer to the actual driving situation of the target vehicle, thereby improving the vehicle recognition accuracy. Description of the Drawings
[0019] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments of the present application. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.
[0020] Figure 1 is a flowchart of a target vehicle recognition method provided by an embodiment of the present application;
[0021] Figure 2 is a schematic diagram of cross - verification provided by an embodiment of the present application;
[0022] Figure 3 is a structural diagram of a target vehicle recognition device provided by an embodiment of the present application;
[0023] Figure 4 is a structural diagram of an electronic device provided by an embodiment of the present application. Specific embodiments
[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts fall within the scope of protection of the present application.
[0025] Please refer to Figure 1 , Figure 1 which is a flowchart of a target vehicle recognition method provided by an embodiment of the present application. As Figure 1 shown, it includes the following steps:
[0026] Step 101: Obtain the picture data and trajectory data of the target vehicle, and a first detection data set. The first detection data set includes the vehicle identifier, detection picture, and detection trajectory corresponding to each of the N vehicles within a first set time. N is a positive integer greater than 1.
[0027] The above - mentioned target vehicle is the vehicle to be recognized. It should be noted that the target vehicle can be a vehicle for which the vehicle identifier has not been recognized, or a vehicle for which the confidence level of the recognition result is lower than the set threshold. Among them, the vehicle identifier is usually the license plate of the vehicle.
[0028] The image data of the target vehicle is the image data of the target object captured by the detection device in the road network, and the trajectory data is the trajectory data of the target object passing by captured by the detection device in the road network. It should be noted that during the process of vehicle recognition, usually the detection devices arranged in the road network recognize each vehicle in the road network. Therefore, for the target vehicle, the recognized image data and trajectory data are the image data and trajectory data captured at the location of the detection device.
[0029] The above first detection data set includes the detection data of the target object. Since the time when the target vehicle is captured and the location of the detection device that captures the target vehicle are known, the possible time and location of the target vehicle in the road network can be determined by setting a certain time range and distance range, and then the detection trajectories and detection images of each vehicle within the time range and distance range are obtained to get the first detection data set.
[0030] Step 102: Identify the vehicles in the first detection data set whose corresponding detection trajectories satisfy the first threshold with the trajectory data, and obtain the first intermediate set.
[0031] It should be noted that among the vehicles in the first detection data set, the detection trajectories of some vehicles are similar to the trajectory data of the target vehicle, while the detection trajectories of some other vehicles are not similar to the trajectory data of the target vehicle. The target vehicle is one of the vehicles with detection trajectories similar to the trajectory data, rather than one of the other vehicles with detection trajectories not similar to the trajectory data. In the embodiments of the present application, the first threshold is used to confirm whether the detection trajectories of the vehicles in the first detection data set are similar to the trajectory data of the target vehicle, and the vehicles that meet the first threshold are identified to obtain the first intermediate set.
[0032] Among them, the vehicles in the first intermediate set are all vehicles in the first detection data set, and the vehicles in the first intermediate set all satisfy that their corresponding detection trajectories satisfy the first threshold with the trajectory data of the target vehicle.
[0033] Step 103: Identify the vehicles in the first intermediate set whose similarity between the corresponding detection images and the image data is greater than the second similarity threshold, and obtain the first set.
[0034] It should be noted that among the detection pictures of some vehicles in the first intermediate set, the picture data is similar to that of the target vehicle, while for the other part of the vehicles, the picture data is not similar to that of the target vehicle. The target vehicle is one of the vehicles in the part of the detection pictures with similar picture data, rather than one of the other part of the vehicles with dissimilar picture data. In the embodiments of the present application, the second similarity threshold is used to confirm whether the detection pictures of the vehicles in the first intermediate set are similar to the picture data of the target vehicle, and the vehicles with a similarity greater than or equal to the second similarity threshold are identified to obtain the first set.
[0035] Among them, the vehicles in the first set are all vehicles in the first intermediate set, and the vehicles in the first set all satisfy that the corresponding detection pictures are greater than or equal to the second similarity threshold compared with the picture data of the target vehicle.
[0036] Specifically, to identify the vehicles in the first intermediate set whose similarity between the corresponding detection pictures and the picture data is greater than the second similarity threshold to obtain the first set, it can be expressed by the following formula:
[0037]
[0038] Among them, is the first set, C i is the first intermediate set, y is the detection picture of the vehicle in the first intermediate set, x is the picture data of the target vehicle, s(x, y) is the similarity between the monitoring picture and the picture data, and α is the second similarity threshold.
[0039] Furthermore, the order of identifying the vehicles similar to the trajectory data and picture data of the target vehicle by the detection trajectory and the detection picture can be adjusted, that is, the first set can also be obtained in the following way:
[0040] Identify the vehicles in the first detection data set whose similarity between the corresponding detection pictures and the picture data is greater than the second similarity threshold to obtain an intermediate set, and then identify the vehicles in the intermediate set whose corresponding detection trajectories satisfy the first threshold with the trajectory data to obtain the first set.
[0041] Step 104: Mark the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle.
[0042] The detection trajectory corresponding to each vehicle in the above-mentioned first set satisfies the first threshold with the trajectory data of the target vehicle, and the similarity between the detection picture corresponding to each vehicle in the first set and the picture data of the target vehicle is greater than or equal to the second similarity threshold. It can be considered that the target vehicle can be any vehicle in the first set. Mark the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle to realize the confirmation of the identifier of the target vehicle.
[0043] Further, in the case where there are multiple vehicles in the first set, mark the vehicle identifier of the vehicle with the highest similarity between the corresponding detected image and the image data as the vehicle identifier of the target vehicle; or, in the case where there are multiple vehicles in the first set, mark the vehicle identifier of the vehicle with the highest matching degree between the corresponding detected trajectory and the trajectory data as the vehicle identifier of the target vehicle.
[0044] In the embodiments of the present application, by obtaining the image data and trajectory data of the target vehicle, and a first detection data set, the first detection data set includes the vehicle identifier, the detected image, and the detected trajectory corresponding to each vehicle among N vehicles within a first set time, where N is a positive integer greater than 1; identify the vehicles in the first detection data set whose corresponding detected trajectories and the trajectory data satisfy a first threshold to obtain a first intermediate set; identify the vehicles in the first intermediate set whose corresponding detected images and the image data have a similarity greater than a second similarity threshold to obtain a first set; mark the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle. In this way, by combining the detected trajectory and the trajectory data, and the detected image and the image data, the first set is screened from the first detection data set, so as to mark the target vehicle by the vehicle identifier of the vehicle in the first set, making the vehicle identifier of the target vehicle closer to the actual driving situation of the target vehicle, thereby improving the vehicle recognition accuracy.
[0045] In one embodiment, after obtaining the image data and trajectory data of the target vehicle, and the first detection data set, the method further includes:
[0046] Obtain a second detection data set, the second detection data set includes the vehicle identifier, the detected image, and the detected trajectory corresponding to each vehicle among M vehicles within a second set time, the second set time is greater than the first set time, and M is a positive integer greater than 1;
[0047] Before marking the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle, the method further includes:
[0048] Identify the vehicles in the second detection data set whose corresponding detected images and the image data have a similarity to a third similarity threshold to obtain a second intermediate set, the third similarity threshold is less than or equal to the second similarity threshold;
[0049] Identify the vehicles in the second intermediate set whose corresponding detected trajectories and the trajectory data satisfy a fourth threshold to obtain a second set, the fourth threshold is greater than or equal to the first threshold;
[0050] Marking the vehicle identifier of one vehicle in the first set as the vehicle identifier of the target vehicle includes:
[0051] Weight the probability that the first vehicle is selected in the first set and the probability that the first vehicle is selected in the second set to obtain the target probability corresponding to the first vehicle, so as to determine the target probabilities of multiple vehicles, where the multiple vehicles are all the vehicles in the union of the first set and the second set, and the first vehicle is one of the multiple vehicles;
[0052] Mark the vehicle identifier corresponding to the vehicle with the maximum target probability among the multiple vehicles as the vehicle identifier of the target vehicle.
[0053] The above-mentioned second detection data set is the detection data including the target object. Since the time when the target vehicle is photographed and the position of the detection device that photographs the target vehicle are known, the possible time and position of the target vehicle in the road network can be determined by setting a certain time range and distance range, and then the detection trajectories and detection pictures of each vehicle within the time range and distance range are obtained to obtain the first detection data set. It should be noted that the time range and / or spatial range corresponding to the second detection data set should be larger than the time range and / or spatial range corresponding to the first detection data set, that is, the second set time is greater than the first set time.
[0054] The above-mentioned second intermediate set is the vehicle for which the similarity between the detection picture corresponding to the second detection data set and the picture data of the target vehicle is greater than or equal to the third similarity threshold, and the above-mentioned second set is the vehicle for which the detection trajectory corresponding to the second intermediate set and the trajectory data of the target vehicle meet the fourth threshold, that is, each vehicle in the second set is similar to the target vehicle, so as to be used to mark the vehicle identifier of the target vehicle.
[0055] It should be noted that in the embodiments of the present application, the vehicle identifiers of the vehicles in the first set and the second set can both mark the target vehicle. To further improve the recognition accuracy, the vehicle identifier of one vehicle in the first set and the second set is determined by combining the first set and the second set. Specifically:
[0056] Weight the probability that the first vehicle is selected in the first set and the probability that the first vehicle is selected in the second set to obtain the target probability corresponding to the first vehicle, so as to determine the target probabilities of multiple vehicles. The multiple vehicles are all the vehicles in the union of the first set and the second set, and the first vehicle is one of the multiple vehicles. For example, the first set includes three vehicles A, B, and C, and the second set includes four vehicles B, C, D, and E. When the first vehicle is vehicle A, the probability that vehicle A is selected in the first set is 1 / 3, and the probability that vehicle A is selected in the second set is 0; while when the first vehicle is B, the probability that vehicle B is selected in the first set is 1 / 3, and the probability that vehicle B is selected in the second set is 1 / 4.
[0057] Mark the vehicle identifier corresponding to the vehicle with the highest target probability among multiple vehicles as the vehicle identifier of the target vehicle. For example, the weight corresponding to the first set is 1, and the weight corresponding to the second set is 0.5. At this time, the target probability of vehicle A is 1 / 3, the target probabilities of vehicle B and vehicle C are both 1 / 3 + 1 / 8, and the target probabilities of vehicle D and vehicle E are 1 / 8. That is, the target probabilities of vehicle B and vehicle C are the highest. Setting the vehicle identifiers of vehicle B and vehicle C as the vehicle identifier of the target vehicle can further improve the accuracy of vehicle recognition.
[0058] Among them, since the condition for obtaining the first set is more stringent (manifested as the third similarity threshold being less than or equal to the second similarity threshold, and the fourth threshold being greater than or equal to the first threshold), the weight corresponding to the first set is set to be greater than the weight of the second set to avoid the interference of individual data in the second set on the whole and further improve the accuracy of vehicle recognition.
[0059] In one embodiment, before weighting the probability that the first vehicle is selected in the first set and the probability that the first vehicle is selected in the second set to obtain the target probability corresponding to the first vehicle, the method further includes:
[0060] Identifying vehicles in the first detection data set whose corresponding detection trajectories satisfy a fifth threshold with the trajectory data, to obtain a third set, where the fifth threshold is greater than or equal to the fourth threshold;
[0061] Identifying vehicles in the second detection data set whose similarity with the picture data of the corresponding detection pictures is greater than or equal to a sixth similarity threshold, to obtain a fourth set, where the sixth similarity threshold is less than or equal to the third similarity threshold;
[0062] The weighting the probability that the first vehicle is selected in the first set and the probability that the first vehicle is selected in the second set to obtain the target probability corresponding to the first vehicle includes:
[0063] Weighting the probability that the first vehicle is selected in the first set, the probability that the first vehicle is selected in the second set, the probability that the first vehicle is selected in the third set, and the probability that the first vehicle is selected in the fourth set to obtain the target probability corresponding to the first vehicle.
[0064] The above-mentioned third set is a set confirmed only by detecting the trajectory and trajectory data. In this set, there are vehicles for which the corresponding detected pictures are similar or dissimilar to the picture data. The above-mentioned fourth set is a set confirmed only by detecting the pictures and picture data. In this set, there are vehicles for which the corresponding detected trajectories do not match the trajectory data. It should be noted that both the third set and the fourth set include the vehicle identifiers corresponding to the target vehicles. The third set and the fourth set are combined with the first set and the second set to determine the identifier of the target vehicle, so as to further improve the recognition accuracy.
[0065] Specifically, the target probability of the vehicle is expressed by the following formula:
[0066]
[0067] where score(c) is the target probability, is the probability of the vehicle in the first set being selected, is the probability of the vehicle in the second set being selected, is the probability of the vehicle in the third set being selected, is the probability of the vehicle in the fourth set being selected, and w1, w2, w3, and w4 are weight coefficients.
[0068] Furthermore, since there is more data for the vehicles in the third set and the fourth set, and there is more obvious irrelevant data, in order to reduce the interference of the irrelevant data, the weight coefficients corresponding to the third set and the fourth set are less than the weight coefficient of the second set. Among them, the weight coefficients of the third set and the fourth set can be the same.
[0069] Specifically, the cross-validation methods of the first set, the second set, the third set, and the fourth set are as Figure 2 shown. The first set is identified by a high threshold (i.e., the first threshold and the second similarity threshold), the second set is identified by a medium threshold (i.e., the third similarity threshold and the fourth threshold), the third set and the fourth set are identified by a low threshold (i.e., the fifth threshold and the sixth similarity threshold), and then the first set, the second set, the third set, and the fourth set are combined to confirm the target probability, so as to obtain the vehicle identifier of the target vehicle.
[0070] In one embodiment, the fifth threshold includes a first sub-time threshold and a first sub-distance threshold. Identifying the vehicles in the first detected data set for which the corresponding detected trajectories satisfy the fifth threshold to obtain the third set includes:
[0071] Obtain road network data, where the road network data includes multiple roads;
[0072] Project the detection trajectories of the vehicles included in the first detection data set onto the multiple roads to obtain a first projection set, where the first projection set includes multiple first projection trajectories, and each first projection trajectory is the projection trajectory corresponding to the detection trajectory of the vehicle included in the first detection data set;
[0073] Based on the first positions corresponding to the trajectory data at K moments, identify the third set in the first projection trajectory set. For the vehicles in the third set, within a first sub-time threshold from the first moment, the distance between the first projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the first sub-distance threshold. The first moment is any one of the K moments, and K is a positive integer greater than 1.
[0074] The detection trajectories in the above first detection data are the satellite positioning trajectories reported by the vehicles to the cloud. During the driving of the vehicles, each vehicle uploads the satellite positioning data to the cloud to implement functions such as vehicle positioning or navigation. The target vehicle also reports the satellite positioning trajectory during the driving process, so that the detection trajectories in the first detection data include the trajectory of the target vehicle.
[0075] It should be noted that there is a difference between the satellite positioning trajectory and the trajectory data of the target vehicle. The trajectory data of the target vehicle is the data obtained by being captured by detection devices, specifically the roads in the road network and the positions on the roads. However, the satellite positioning trajectory is not directly located in the road network, and it is impossible to directly compare the detection trajectories in the first detection data with the trajectory data of the target vehicle. Therefore, in the embodiments of the present application, the detection trajectories of the vehicles included in the first detection data set are projected onto multiple roads to obtain the first projection set.
[0076] The specific process is as follows:
[0077] 1) First, obtain the road network data. Let the road network data be H = (V, E), where V is the set of intersections in the road network and E is the set of each road. For each road e, the identifier of the road is id(e), and e ∈ E. For the road e, the longitude and latitude of its starting point are (x s , y s ), and the longitude and latitude of its ending point are (x e , y e ). The length of the road e is:
[0078]
[0079] L e is the length of the road e.
[0080] 2) Project the detection trajectories of the vehicles included in the first detection data set onto multiple roads. Let the satellite positioning trajectory be G = (g1, g2,..., gk ,), where g i is a trajectory point in trajectory G, g i =(x i , y i , t i ), x i is the longitude of trajectory point g i , y i is the latitude of trajectory point g i , and t i is the time of trajectory point g i . Let the longitude and latitude of a trajectory point f on road e in the detected trajectory be (x f , y f ), then the direction vector from trajectory point f to the starting point of road e is (x f - x s , y f - y s ), and the direction vector of road e is (x e - x s , y e - y s ). Then the projection of trajectory point f on road e is:
[0081]
[0082] P f is the projection of trajectory point f on road e. Through the above formula, the first projected trajectory of the detected trajectory projected on the road network can be obtained, and then the first projection set can be obtained.
[0083] Furthermore, after obtaining the first projection set, it is determined whether each first projected trajectory in the first projection set satisfies the first threshold with the trajectory data of the target vehicle, that is, within the first sub - time threshold from the first moment, the distance between the first projected trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the first sub - distance threshold.
[0084] Specifically, let the first projected trajectory be R=(r1, r2,..., r k ), r i =(id(e ri ), p r ), the trajectory data of the target vehicle be Q=(q1, q2,..., q k ), q i =(id(e qi ), p q ), where id(e ri ) is the identifier of the road where trajectory point r i is located, id(e qi ) is the identifier of the road where trajectory point q iThe identifier of the road where it is located, p r is the trajectory point r i at the position of p q is the trajectory point q i at the position of. The first set can be represented by the following formula:
[0085] C i ={c j |id(e ri ) = id(e qi ), |p r - p q | ≤ σ, |t j - t i | ≤ τ}
[0086] C i is the first set, c j is the vehicle identifier corresponding to the trajectory point j in the first projected trajectory, t j is the time of the trajectory point j in the first projected trajectory, t i is the time of the trajectory data, σ is the first sub - distance threshold, and τ is the first sub - time threshold.
[0087] Optionally, the first sub - distance threshold is the first sub - relative position ratio threshold. The distance between the first projected trajectory corresponding to the vehicle and the first position corresponding to the trajectory data at the first moment is less than or equal to the first sub - distance threshold. Specifically, the difference between the relative position ratio of the first projected trajectory corresponding to the vehicle on the road and the relative position ratio of the trajectory data on the road is less than or equal to the first sub - relative position ratio threshold. Among them, the position of the first projected trajectory in the road network is P f , and the relative position ratio is p f = P f / L e .
[0088] Optionally, based on the first positions corresponding to the trajectory data at K moments, a third set is identified in the first projected trajectory set. Among them, the K moments are the moments when the target vehicle passes through the intersections in the road network. The duration between the moment when the first projected trajectory in the third set passes through the intersection and the moment when the target vehicle passes through the intersection is less than the duration threshold. It should be noted that in this optional embodiment, by comparing the moment when the first projected trajectory passes through the intersection with the moment when the target vehicle passes through the intersection and combining the duration threshold screening, the third set is obtained.
[0089] Among them, the first projected trajectory passes through the trajectory point g1, the trajectory point g2, and the intersection in sequence. The moment of passing through the intersection is calculated by the following formula:
[0090]
[0091] t1 is the time at the trajectory point g1 passed, t2 is the time at the trajectory point g2 passed, and t s is the time passing through the intersection, p1 is the position of the trajectory point g1 on the road, p2 is the position of the trajectory point g2 on the road, and q is the position of the intersection.
[0092] In one embodiment, the fourth threshold includes a second sub - time threshold and a second sub - distance threshold. Identifying the vehicles in the second intermediate set whose corresponding detected trajectories and the trajectory data satisfy the fourth threshold to obtain a second set includes:
[0093] Obtain road network data, where the road network data includes multiple roads;
[0094] Project the detected trajectories of the vehicles included in the second intermediate set onto the multiple roads to obtain a second projection set. The second projection set includes multiple second projection trajectories, and each second projection trajectory is the projection trajectory corresponding to the detected trajectory of the vehicle included in the second intermediate set;
[0095] Based on the first positions corresponding to the K moments of the trajectory data, identify in the second projection trajectory set to obtain the second set. For the vehicles in the second set, within the second sub - time threshold from the first moment, the distance between the second projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the second sub - distance threshold, and the first moment is any one of the K moments.
[0096] It should be noted that the specific process of identifying the vehicles in the second intermediate set whose corresponding detected trajectories and the trajectory data satisfy the fourth threshold to obtain the second set is the same as that of identifying the vehicles in the first detected data set whose corresponding detected trajectories and the trajectory data satisfy the fifth threshold to obtain the third set, and will not be elaborated here.
[0097] In the embodiments of the present application, by obtaining road network data, where the road network data includes multiple roads; projecting the detected trajectories of the vehicles included in the second intermediate set onto the multiple roads to obtain a second projection set, and then identifying based on the first positions corresponding to the K moments of the trajectory data in the second projection trajectory set to achieve obtaining the second set.
[0098] In one embodiment, the first threshold includes a third sub - time threshold and a third sub - distance threshold; identifying the vehicles in the first detected data set whose corresponding detected trajectories and the trajectory data satisfy the first threshold to obtain a first intermediate set includes:
[0099] Obtain road network data, where the road network data includes multiple roads;
[0100] Project the detection trajectories of the vehicles included in the first detection data set onto the multiple roads to obtain a first projection set, where the first projection set includes multiple first projection trajectories, and each first projection trajectory is the projection trajectory corresponding to the detection trajectory of the vehicle included in the first detection data set;
[0101] Based on the first positions corresponding to the trajectory data at K moments, identify the first intermediate set in the first projection trajectory set. For the vehicles in the first intermediate set, within a third sub-time threshold from the first moment, the distance between the first projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the third sub-distance threshold, and the first moment is any one of the K moments.
[0102] It should be noted that the specific processes of identifying the vehicles whose corresponding detection trajectories in the first detection data set satisfy the first threshold to obtain the first intermediate set and the second set, and identifying the vehicles whose corresponding detection trajectories in the first detection data set satisfy the fifth threshold to obtain the third set are the same, and will not be elaborated here.
[0103] In the embodiments of the present application, by obtaining road network data, where the road network data includes multiple roads; projecting the detection trajectories of the vehicles included in the first detection data set onto the multiple roads to obtain a first projection set, and then identifying based on the first positions corresponding to the trajectory data at K moments in the first projection trajectory set to achieve obtaining the first intermediate set.
[0104] In one embodiment, the identifying the second set from the second intermediate set based on the first positions corresponding to the K moments includes:
[0105] In the case that the second projection set includes the second projection trajectory of the second vehicle within a second sub-time threshold from the first moment, based on whether the distance between the second projection trajectory of the second vehicle and the first position corresponding to the first moment is less than or equal to the second sub-distance threshold, confirm whether the second vehicle is a vehicle in the second set to obtain the second set, where the second vehicle is any vehicle in the second intermediate set;
[0106] In the case that the second projection set does not include the second projection trajectory of the second vehicle within a second sub-time threshold from the first moment, obtain the historical trajectory data of the second vehicle;
[0107] Construct a trajectory model of the second vehicle based on the historical trajectory data of the second vehicle;
[0108] Calculate a third position corresponding to each of the K moments based on the trajectory model of the second vehicle;
[0109] Based on whether the distance between the third position and the first position at each of the moments is greater than the second sub-distance threshold, determine whether the second vehicle is a vehicle in the second set, so as to obtain the second set.
[0110] It should be noted that in the second detection data set, there is a second vehicle whose detection picture and the picture data of the target vehicle meet the set similarity threshold. However, during the time when the picture of the target vehicle is taken, the second vehicle does not upload its trajectory, that is, there is no detection trajectory of the second vehicle in the second detection data set. To avoid misidentification caused by abnormal non-upload of the vehicle, in this case, the detection trajectory corresponding to the second vehicle is determined through the historical trajectory data of the second vehicle, so as to further improve the accuracy of vehicle identification.
[0111] Among them, the trajectory model of the second vehicle is obtained by training the initial model with the historical trajectory data of the second vehicle and the time corresponding to the historical trajectory data. Calculate the third position corresponding to each of the K moments through the trajectory model of the second vehicle to obtain the detection trajectory corresponding to the second vehicle.
[0112] Further, in the process of calculating the third position corresponding to each of the K moments through the trajectory model of the second vehicle, calculate the confidence level of each third position. When the confidence level is lower than the set confidence level threshold, it is considered that the calculated third position of the second vehicle is not feasible, and it is determined that the second vehicle is not a vehicle in the second set.
[0113] Please refer to Figure 3 , Figure 3 which is a structural diagram of a target vehicle recognition device provided by an embodiment of the present application. As Figure 3 shown, the target vehicle recognition device 300 includes:
[0114] A first acquisition module 301, configured to acquire picture data and trajectory data of a target vehicle, and a first detection data set, where the first detection data set includes vehicle identifiers, detection pictures, and detection trajectories corresponding to each of N vehicles within a first set time, and N is a positive integer greater than 1;
[0115] A first recognition module 302, configured to recognize vehicles in the first detection data set whose corresponding detection trajectories satisfy a first threshold with the trajectory data, to obtain a first intermediate set;
[0116] A second recognition module 303, configured to recognize vehicles in the first intermediate set whose similarity between the corresponding detection pictures and the picture data is greater than a second similarity threshold, to obtain a first set;
[0117] A marking module 304 for marking the vehicle identifier of a vehicle in the first set as the vehicle identifier of the target vehicle.
[0118] In one embodiment, after the first acquisition module 301, the target vehicle recognition device 300 further includes:
[0119] A second acquisition module for acquiring a second set of detection data, where the second set of detection data includes the vehicle identifier, detection picture, and detection trajectory corresponding to each vehicle among M vehicles within a second set time, the second set time being greater than the first set time, and M being a positive integer greater than 1;
[0120] Before the marking module 304, the target vehicle recognition device 300 further includes:
[0121] A third recognition module for recognizing vehicles in the second set of detection data whose similarity of the corresponding detection pictures to the picture data is less than or equal to a third similarity threshold, to obtain a second intermediate set, where the third similarity threshold is less than or equal to the second similarity threshold;
[0122] A fourth recognition module for recognizing vehicles in the second intermediate set whose corresponding detection trajectories satisfy a fourth threshold with respect to the trajectory data, to obtain a second set, where the fourth threshold is greater than or equal to the first threshold;
[0123] The marking module 304 includes:
[0124] A first processing sub-module for weighting the probability that a first vehicle is selected in the first set and the probability that the first vehicle is selected in the second set, to obtain a target probability corresponding to the first vehicle, so as to determine the target probabilities of multiple vehicles, where the multiple vehicles are all vehicles in the union of the first set and the second set, and the first vehicle is one of the multiple vehicles;
[0125] A marking sub-module for marking the vehicle identifier corresponding to the vehicle with the maximum target probability among the multiple vehicles as the vehicle identifier of the target vehicle.
[0126] In one embodiment, before the first processing sub-module, the target vehicle recognition device 300 further includes:
[0127] A fifth recognition module for recognizing vehicles in the first set of detection data whose corresponding detection trajectories satisfy a fifth threshold with respect to the trajectory data, to obtain a third set, where the fifth threshold is greater than or equal to the fourth threshold;
[0128] The sixth recognition module is configured to recognize vehicles in the second detection data set whose similarity between the corresponding detected pictures and the picture data is greater than or equal to a sixth similarity threshold, to obtain a fourth set, where the sixth similarity threshold is less than or equal to the third similarity threshold;
[0129] The first processing sub-module includes:
[0130] The first processing unit is configured to weight the probability that the first vehicle is selected in the first set, the probability that the first vehicle is selected in the second set, the probability that the first vehicle is selected in the third set, and the probability that the first vehicle is selected in the fourth set, to obtain the target probability corresponding to the first vehicle.
[0131] In one embodiment, the fifth threshold includes a first sub-time threshold and a first sub-distance threshold, and the fifth recognition module includes:
[0132] The first acquisition sub-module is configured to acquire road network data, where the road network data includes a plurality of roads;
[0133] The second processing sub-module is configured to project the detection trajectories of the vehicles included in the first detection data set onto the plurality of roads, to obtain a first projection set, where the first projection set includes a plurality of first projection trajectories, and each first projection trajectory is a projection trajectory corresponding to the detection trajectory of the vehicle included in the first detection data set;
[0134] The first recognition sub-module is configured to, based on the first positions corresponding to the trajectory data at K moments, recognize the third set in the first projection trajectory set, where for the vehicles in the third set, within a first sub-time threshold from the first moment, the distance between the first projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the first sub-distance threshold, the first moment is any one of the K moments, and K is a positive integer greater than 1.
[0135] In one embodiment, the fourth threshold includes a second sub-time threshold and a second sub-distance threshold, and the fourth recognition module includes:
[0136] The second acquisition sub-module is configured to acquire road network data, where the road network data includes a plurality of roads;
[0137] The third processing sub-module is configured to project the detection trajectories of the vehicles included in the second intermediate set onto the plurality of roads, to obtain a second projection set, where the second projection set includes a plurality of second projection trajectories, and each second projection trajectory is a projection trajectory corresponding to the detection trajectory of the vehicle included in the second intermediate set;
[0138] A second recognition sub-module, configured to recognize the second set from the second projected trajectory set based on the first positions corresponding to the trajectory data at K moments, where, within a second sub-time threshold from the first moment for the vehicles in the second set, the distance between the second projected trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the second sub-distance threshold, and the first moment is any one of the K moments.
[0139] In one embodiment, the second recognition sub-module includes:
[0140] A first recognition unit, configured to, when the second projected set includes the second projected trajectory of a second vehicle within a second sub-time threshold from the first moment, confirm whether the second vehicle is a vehicle in the second set based on whether the distance between the second projected trajectory of the second vehicle and the first position corresponding to the first moment is less than or equal to the second sub-distance threshold, so as to obtain the second set, where the second vehicle is any one of the vehicles in the second intermediate set;
[0141] An acquisition unit, configured to, when the second projected set does not include the second projected trajectory of the second vehicle within a second sub-time threshold from the first moment, acquire the historical trajectory data of the second vehicle;
[0142] A second processing unit, configured to construct a trajectory model of the second vehicle based on the historical trajectory data of the second vehicle;
[0143] A third processing unit, configured to calculate the third position corresponding to each of the K moments based on the trajectory model of the second vehicle;
[0144] A second recognition module, configured to determine whether the second vehicle is a vehicle in the second set based on whether the distance between the third position and the first position at each moment is greater than the second sub-distance threshold, so as to obtain the second set.
[0145] In one embodiment, the first threshold includes a third sub-time threshold and a third sub-distance threshold; the first recognition module 302 includes:
[0146] A third acquisition sub-module, configured to acquire road network data, where the road network data includes a plurality of roads;
[0147] A fourth processing sub-module, configured to project the detection trajectories of the vehicles included in the first detection data set onto the plurality of roads, to obtain a first projected set, where the first projected set includes a plurality of first projected trajectories, and each first projected trajectory is a projected trajectory corresponding to the detection trajectory of the vehicle included in the first detection data set;
[0148] A third recognition sub-module, configured to identify the first intermediate set from the first projection trajectory set based on the first positions corresponding to the trajectory data at K moments, where, within a third sub-time threshold from the first moment in the first intermediate set, the distance between the first projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the third sub-distance threshold, and the first moment is any one of the K moments.
[0149] The target vehicle recognition device provided in the embodiments of the present application can implement each process of the above-mentioned target vehicle recognition method. The technical features correspond one by one and can achieve the same technical effects. To avoid repetition, they will not be elaborated here.
[0150] It should be noted that the target vehicle recognition device in the embodiments of the present application can be a device, or a component, an integrated circuit, or a chip in an electronic device.
[0151] The embodiments of the present application further provide an electronic device. Refer to Figure 4 , Figure 4 which is a schematic structural diagram of an electronic device provided in the embodiments of the present application. The electronic device includes a memory 401, a processor 402, and a program or instruction running on the memory 401. When the program or instruction is executed by the processor 402, it can implement Figure 1 any step in the corresponding method embodiment and achieve the same beneficial effects. Details are not described herein again.
[0152] Among them, the processor 402 can be a CPU, an ASIC, an FPGA, or a GPU.
[0153] Those of ordinary skill in the art can understand that all or part of the steps for implementing the method in the above embodiments can be completed by hardware related to program instructions, and the program can be stored in a readable medium.
[0154] The embodiments of the present application further provide a readable storage medium. A computer program is stored on the readable storage medium. When the computer program is executed by a processor, it can implement any step in the corresponding method embodiment described above Figure 1 and can achieve the same technical effects. To avoid repetition, they will not be elaborated here. The storage medium can be, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.
[0155] The terms "first", "second", etc. in the embodiments of the present application are used to distinguish similar objects and do not necessarily describe a specific order or sequence. 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 limit to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices. In addition, the use of "and / or" in the present application means at least one of the connected objects. For example, A and / or B and / or C means including the 7 cases of A alone, B alone, C alone, A and B existing together, B and C existing together, A and C existing together, and A, B, and C existing together.
[0156] It should be noted that in this text, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not clearly listed, or elements inherent to such a process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the existence of additional identical elements in the process, method, article or device comprising that element.
[0157] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods 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. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to enable a terminal (which can be a mobile phone, computer, server, air conditioner, or a second terminal device, etc.) to execute the methods of the various embodiments of the present application.
[0158] The above describes the embodiments of the present application in conjunction with the drawings, but the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all belong to the protection scope of the present application.
Claims
1. A method for identifying a target vehicle, characterized in that Including: Obtain the image data and trajectory data of the target vehicle, as well as a first detection data set, where the first detection data set includes the vehicle identifier, detection image, and detection trajectory corresponding to each of N vehicles within a first set time, and N is a positive integer greater than 1; Identify the vehicles in the first detection data set whose corresponding detection trajectories satisfy a first threshold with the trajectory data to obtain a first intermediate set; Identify the vehicles in the first intermediate set whose similarity of the corresponding detection images to the image data is greater than a second similarity threshold to obtain a first set; Mark the vehicle identifier of a vehicle in the first set as the vehicle identifier of the target vehicle.
2. The method according to claim 1, characterized in that, After obtaining the image data and trajectory data of the target vehicle, as well as the first detection data set, the method further includes: Obtain a second detection data set, where the second detection data set includes the vehicle identifier, detection image, and detection trajectory corresponding to each of M vehicles within a second set time, the second set time is greater than the first set time, and M is a positive integer greater than 1; Before marking the vehicle identifier of a vehicle in the first set as the vehicle identifier of the target vehicle, the method further includes: Identify the vehicles in the second detection data set whose similarity of the corresponding detection images to the image data is a third similarity threshold to obtain a second intermediate set, where the third similarity threshold is less than or equal to the second similarity threshold; Identify the vehicles in the second intermediate set whose corresponding detection trajectories satisfy a fourth threshold with the trajectory data to obtain a second set, where the fourth threshold is greater than or equal to the first threshold; The step of marking the vehicle identifier of a vehicle in the first set as the vehicle identifier of the target vehicle includes: Weight the probability of a first vehicle being selected in the first set and the probability of the first vehicle being selected in the second set to obtain the target probability corresponding to the first vehicle, so as to determine the target probabilities of multiple vehicles, where the multiple vehicles are all the vehicles in the union of the first set and the second set, and the first vehicle is one of the multiple vehicles; Mark the vehicle identifier of the vehicle corresponding to the maximum target probability among the multiple vehicles as the vehicle identifier of the target vehicle.
3. The method according to claim 2, characterized in that, Before weighting the probability of a first vehicle being selected in the first set and the probability of the first vehicle being selected in the second set to obtain the target probability corresponding to the first vehicle, the method further includes: Identify the vehicles in the first detection data set whose corresponding detection trajectories satisfy a fifth threshold with the trajectory data to obtain a third set, where the fifth threshold is greater than or equal to the fourth threshold; Identify the vehicles in the second detection data set whose similarity of the corresponding detection images to the image data is greater than or equal to a sixth similarity threshold to obtain a fourth set, where the sixth similarity threshold is less than or equal to the third similarity threshold; The step of weighting the probability of a first vehicle being selected in the first set and the probability of the first vehicle being selected in the second set to obtain the target probability corresponding to the first vehicle includes: Weight the probability that the first vehicle is selected in the first set, the probability that the first vehicle is selected in the second set, the probability that the first vehicle is selected in the third set, and the probability that the first vehicle is selected in the fourth set to obtain the target probability corresponding to the first vehicle.
4. The method according to claim 3, wherein The fifth threshold includes a first sub-time threshold and a first sub-distance threshold. Identifying the vehicles in the first detection data set whose corresponding detection trajectories satisfy the fifth threshold with the trajectory data to obtain the third set includes: Obtain road network data, where the road network data includes multiple roads; Project the detection trajectories of the vehicles included in the first detection data set onto the multiple roads to obtain a first projection set, where the first projection set includes multiple first projection trajectories, and each first projection trajectory is the projection trajectory corresponding to the detection trajectory of the vehicle included in the first detection data set; Based on the first positions corresponding to the K moments of the trajectory data, identify the third set in the first projection trajectory set. For the vehicles in the third set, within the first sub-time threshold from the first moment, the distance between the first projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the first sub-distance threshold. The first moment is any one of the K moments, and K is a positive integer greater than 1.
5. The method according to claim 2, wherein The fourth threshold includes a second sub-time threshold and a second sub-distance threshold. Identifying the vehicles in the second intermediate set whose corresponding detection trajectories satisfy the fourth threshold with the trajectory data to obtain the second set includes: Obtain road network data, where the road network data includes multiple roads; Project the detection trajectories of the vehicles included in the second intermediate set onto the multiple roads to obtain a second projection set, where the second projection set includes multiple second projection trajectories, and each second projection trajectory is the projection trajectory corresponding to the detection trajectory of the vehicle included in the second intermediate set; Based on the first positions corresponding to the K moments of the trajectory data, identify the second set in the second projection trajectory set. For the vehicles in the second set, within the second sub-time threshold from the first moment, the distance between the second projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the second sub-distance threshold. The first moment is any one of the K moments.
6. The method according to claim 5, characterized in that, The identifying the second set in the second projection trajectory set based on the first positions corresponding to the K moments of the trajectory data includes: When the second projection trajectory of the second vehicle is included in the second projection set within the second sub-time threshold from the first moment, confirm whether the second vehicle is a vehicle in the second set based on whether the distance between the second projection trajectory of the second vehicle and the first position corresponding to the first moment is less than or equal to the second sub-distance threshold to obtain the second set. The second vehicle is any vehicle in the second intermediate set; In the case that the second projection set does not include the second projection trajectory of the second vehicle within the second sub-time threshold from the first moment, obtain the historical trajectory data of the second vehicle; Construct a trajectory model of the second vehicle based on the historical trajectory data of the second vehicle; Calculate the third position corresponding to each of the K moments based on the trajectory model of the second vehicle; Determine whether the second vehicle is a vehicle in the second set based on whether the distance between the third position and the first position at each moment is greater than the second sub-distance threshold, so as to obtain the second set.
7. The method according to claim 1, wherein The first threshold includes a third sub-time threshold and a third sub-distance threshold; identifying the vehicles in the first detection data set whose corresponding detection trajectories and the trajectory data satisfy the first threshold to obtain a first intermediate set, including: Obtain road network data, where the road network data includes multiple roads; Project the detection trajectories of the vehicles included in the first detection data set onto the multiple roads to obtain a first projection set, where the first projection set includes multiple first projection trajectories, and each first projection trajectory is the projection trajectory corresponding to the detection trajectory of the vehicle included in the first detection data set; Based on the first positions corresponding to the trajectory data at K moments, identify the first intermediate set in the first projection trajectory set. For the vehicles in the first intermediate set, within the third sub-time threshold from the first moment, the distance between the first projection trajectory corresponding to the vehicle and the first position corresponding to the first moment is less than or equal to the third sub-distance threshold, and the first moment is any one of the K moments.
8. An object vehicle recognition device, characterized in that, Including: A first acquisition module, configured to acquire picture data and trajectory data of a target vehicle, and a first detection data set, where the first detection data set includes a vehicle identifier, a detection picture, and a detection trajectory corresponding to each of N vehicles within a first set time, and N is a positive integer greater than 1; A first identification module, configured to identify the vehicles in the first detection data set whose corresponding detection trajectories and the trajectory data satisfy the first threshold to obtain a first intermediate set; A second identification module, configured to identify the vehicles in the first intermediate set whose similarity between the corresponding detection pictures and the picture data is greater than a second similarity threshold to obtain a first set; A marking module, configured to mark the vehicle identifier of a vehicle in the first set as the vehicle identifier of the target vehicle.
9. An electronic device, characterized in that, Including a processor, a memory, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, the steps in the target vehicle identification method according to any one of claims 1 to 7 are implemented.
10. A readable storage medium for storing a program, characterized in that, When the program is executed by the processor, the steps in the target vehicle identification method according to any one of claims 1 to 7 are implemented.