技术领域technical field
本发明涉及智能交通技术领域，具体为一种基于车辆检索的违章图片自动审核方法及系统。The invention relates to the technical field of intelligent transportation, in particular to a method and system for automatically reviewing illegal pictures based on vehicle retrieval.
背景技术Background technique
近年来，随着我国车辆的普及，城市车辆数量逐年增加，交通违章事件也逐年增加。车辆违章事件的查处有利于道路交通规范的建立，可以维护道路交通秩序，预防和减少交通事故，保护人身安全，保护人民的财产安全。所以需要对交通违章事件进行查处。In recent years, with the popularization of vehicles in our country, the number of urban vehicles has increased year by year, and the incidents of traffic violations have also increased year by year. The investigation and handling of vehicle violations is conducive to the establishment of road traffic regulations, can maintain road traffic order, prevent and reduce traffic accidents, protect personal safety, and protect people's property safety. Therefore, it is necessary to investigate and deal with traffic violations.
随着社会的进步和发展，交通违章事件的查处方式也发生了变化，从最初的认为处理，到现在的自动化处理。随着交通违章事件的增加，极大的增加了交通管理部门的人力资源消耗。为了释放因交通违章事件增加的人力资源消耗，出现了违章自动审核系统。With the progress and development of society, the way of investigation and handling of traffic violations has also changed, from the initial thought process to the current automatic process. With the increase of traffic violation incidents, the human resource consumption of the traffic management department has been greatly increased. In order to release the consumption of human resources increased by traffic violations, an automatic review system for traffic violations has emerged.
当前已有许多学者提出了不同的车辆违章图片审核系统，但是交管摄像头所拍摄的视频经过视频分析软件后，抓拍了其所认为的违章车辆图片，但是由于视频分析软件存在一定的误检，给违章审核工作带来一定的困扰。At present, many scholars have proposed different vehicle violation picture review systems. However, after the video captured by the traffic control camera passes through the video analysis software, it captures the picture of the vehicle that it thinks is illegal. Violation review work brings certain troubles.
发明内容Contents of the invention
为克服现有技术中存在的不足，本发明的目的在于提供基于车辆检索的违章图片自动审核方法及系统。In order to overcome the deficiencies in the prior art, the object of the present invention is to provide a method and system for automatically reviewing illegal pictures based on vehicle retrieval.
所述的一种基于车辆检索的违章图片自动审核系统，其特征在于由车辆检测模块、车辆特征提取模块、车辆对比模块及车辆违章检测模块构成，所述车辆检测模块、车辆对比模块分别与车辆特征提取模块、车辆违章检测模块连接，车辆检测模块用于检测出违章图片中所有车辆；车辆特征提取模块用于将车辆检测模块检测出来的车辆进行车辆深度特征提取；车辆对比模块用于将系统判断的目标违章车辆的车辆特征与违章图片中车辆特征提取模块提取到车辆的车辆深度特征进行比对，计算出车辆特征匹配度，并将计算结果发送给车辆违章检测模块；车辆违章检测模块收到车辆对比模块的结果，并根据车辆检测模块提取的相应车辆的坐标，用于对违章车辆中与目标违章车辆的相同车辆定位，再根据具体定位确定违章车辆。Described a kind of automatic inspection system for illegal pictures based on vehicle retrieval is characterized in that it is composed of a vehicle detection module, a vehicle feature extraction module, a vehicle comparison module and a vehicle violation detection module, and the vehicle detection module and the vehicle comparison module are respectively connected with the vehicle The feature extraction module and the vehicle violation detection module are connected. The vehicle detection module is used to detect all vehicles in the violation picture; the vehicle feature extraction module is used to extract the vehicle depth features from the vehicles detected by the vehicle detection module; the vehicle comparison module is used to integrate the system Compare the vehicle features of the judged target violating vehicle with the vehicle depth features extracted by the vehicle feature extraction module in the violation picture, calculate the vehicle feature matching degree, and send the calculation result to the vehicle violation detection module; the vehicle violation detection module receives According to the results of the vehicle comparison module, and according to the coordinates of the corresponding vehicles extracted by the vehicle detection module, it is used to locate the same vehicle as the target illegal vehicle among the illegal vehicles, and then determine the illegal vehicle according to the specific positioning.
基于所述车辆检索的违章图片自动审核系统的车辆审核方法，其特征在于包括如下步骤：The vehicle examination method based on the automatic examination system of illegal picture of described vehicle retrieval, it is characterized in that comprising the following steps:
1)将违章图片以图片中心点像素分为四个区域，分别为第一、第二、第三及第四区域，系统检测出的目标违章车辆存在于第四区域，令车辆检测模块从违章图片的四个区域中提取出同张图片中所有车辆特征向量集合并检测出所有车辆与车辆特征提取模块提取到所有车辆的车辆特征向量的映射集合为其中cj表示第j辆车，n表示车辆数量，表示第j辆车的特征向量的集合，表示第j辆车的车辆特征集合中第k个车辆特征向量；1) Divide the illegal image into four areas by the pixels of the center point of the image, which are the first, second, third, and fourth areas. The target illegal vehicle detected by the system exists in the fourth area, so that the vehicle detection module starts from the illegal Extract all vehicle feature vector sets in the same picture from the four regions of the picture And detect all vehicles and the mapping set of vehicle feature vectors extracted by the vehicle feature extraction module to all vehicles is where cj represents the jth vehicle, n represents the number of vehicles, Represents the set of feature vectors of the jth car, Represents the kth vehicle feature vector in the vehicle feature set of the jth vehicle;
2)由车辆特征提取模块提取出第四区域目标违章车辆的特征向量Fm＝{feati|i＝1,2,...,4096}，i表示第i个车辆特征向量，并将目标违章车辆的特征向量信息传送给车辆对比模块；2) The feature vector Fm ={feati |i=1,2,...,4096} of the target illegal vehicle in the fourth area is extracted by the vehicle feature extraction module, where i represents the i-th vehicle feature vector, and the target The feature vector information of the violating vehicle is sent to the vehicle comparison module;
3)车辆对比模块将提取到的目标违章车辆的车辆特征向量根据公式(1)与步骤1)中提取出的同张图片中所有车辆特征向量集合中的每个车辆特征向量的相似度s,得到相似度集合若两辆车的相似度s＞K0，K0为车辆置信度阈值，则将此车辆信息(cn,carlistn)加入相似车辆集合，cn表示车辆具体编号，carlistn表示车辆所在位置的具体框，由此获得所有违章目标车辆的相似车辆；3) The vehicle comparison module extracts the vehicle feature vector of the target illegal vehicle according to the set of all vehicle feature vectors in the same picture extracted in formula (1) and step 1) Each vehicle feature vector in The similarity s, get the similarity set If the similarity of two vehicles s>K0 , K0 is the vehicle confidence threshold, then add the vehicle information (cn , carlistn ) to the set of similar vehicles, cn indicates the specific number of the vehicle, and carlistn indicates the location of the vehicle The specific box of , thereby obtaining similar vehicles of all violation target vehicles;
其中Q，P分别代表两辆车的特征向量，Q＝{qi|i＝1,2,...,4096}，P＝{pi|i＝1,2,...,4096}，qi表示特征向量Q中的第i个特征值，Pi表示特征向量P中的第i个特征值，K表示车辆特征维度，s表示Q,P两辆车的相似度函数；Where Q and P represent the feature vectors of the two vehicles respectively, Q={qi |i=1,2,...,4096}, P={pi |i=1,2,...,4096} , qi represents the i-th eigenvalue in the eigenvector Q, Pi represents the i-th eigenvalue in the eigenvector P, K represents the vehicle feature dimension, s represents the similarity function of Q and P two vehicles;
4)根据步骤3)中所获得的违章目标车辆的相似车辆集合s，获得相似车辆所在位置的具体框，根据公式(2)获得车辆具体框的中心点坐标，确定相似车辆所在的位置，得到相似车辆中心点坐标(Xc,Yc)；4) According to the similar vehicle set s of the illegal target vehicle obtained in step 3), obtain the specific frame of the location of the similar vehicle, obtain the coordinates of the center point of the specific frame of the vehicle according to the formula (2), determine the location of the similar vehicle, and obtain Similar vehicle center point coordinates (Xc , Yc );
其中，(Xc,Yc)是相似车辆中心点坐标，(X,Y)表示相似车辆具体框左顶点坐标，W表示具体框的宽，H表示具体框的高；Wherein, (Xc , Yc ) are the coordinates of the center point of similar vehicles, (X, Y) represent the coordinates of the left apex of the specific frame of similar vehicles, W represents the width of the specific frame, and H represents the height of the specific frame;
车辆违章检测模块根据相似车辆中心点坐标判断车辆所在的具体区域，获得每个区域相似车辆集合S1＝{(cx,carlistx)|x＝1,2,...,n}，cx表示相似车辆的具体编号，令D＝max(S1)，获得D所在的相似车辆的具体框就是当前区域的目标违章车辆，判断第一区域、第二区域及第三区域中是否都存在目标违章车辆，如若都存在，则目标违章车辆确定违章，反之则为误检结果。The vehicle violation detection module judges the specific area where the vehicle is located according to the coordinates of the center point of similar vehicles, and obtains a set of similar vehicles in each area S1 ={(cx ,carlistx )|x=1,2,...,n}, cx represents the specific serial number of the similar vehicle, let D=max(S1 ), the specific frame of the similar vehicle where D is obtained is the target illegal vehicle in the current area, and it is judged whether there are any in the first area, the second area and the third area If the target violating vehicles exist, the target violating vehicles are determined to be violating the rules, otherwise it is a false detection result.
通过采用上述技术，与现有技术相比，本发明的有益效果如下：By adopting above-mentioned technology, compared with prior art, the beneficial effect of the present invention is as follows:
1)本系统可以在视频分析软件判断车辆违章误检时，得到正确的判断，避免车辆违章的误判，使车辆违章自动化更加完善；1) This system can get the correct judgment when the video analysis software judges the misjudgment of vehicle violations, avoiding misjudgments of vehicle violations, and making the automation of vehicle violations more perfect;
2)本系统可以自动识别车辆是否发生违章事件，释放了交管人力资源，提高了资源利用率。2) This system can automatically identify whether a vehicle has violated the regulations, which releases the human resources of traffic control and improves the utilization rate of resources.
附图说明Description of drawings
图1为本发明方法的步骤流程图；Fig. 1 is the flow chart of the steps of the inventive method;
图2为本发明实施例的检测实例灰度图，其中第四区域的车辆为系统识别到的目标违章车辆；Fig. 2 is the grayscale image of the detection example of the embodiment of the present invention, wherein the vehicle in the fourth area is the target illegal vehicle identified by the system;
图3为本发明审核过程灰度图，在检测实例图的四个区域内找到的与目标车辆的相同车辆。Fig. 3 is a grayscale image of the audit process of the present invention, and the same vehicle as the target vehicle is found in the four areas of the detection example image.
具体实施方式Detailed ways
下面结合说明书附图和实施例，对本发明进行进一步的说明。应当理解，此处所描述的具体实施例仅仅用于解释本发明，并不用于限定本发明。The present invention will be further described below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.
如图1所示，本发明的一种基于车辆检索的违章图片自动审核系统，由车辆检测模块1、车辆特征提取模块2、车辆对比模块3及车辆违章检测模块4构成，所述车辆检测模块1、车辆对比模块3、车辆违章检测模块4与车辆检测模块1和车辆对比模块3分别连接，车辆检测模块1负责对违章图片进行车辆检测，违章图片的分辨率为7296*5792，将违章图片中所有车辆都检测出来；车辆特征提取模块2用于将车辆检测模块1检测出来的车辆进行车辆深度特征提取；车辆对比模块3计算车辆特征匹配度，用于将系统判断的目标违章车辆的车辆特征与违章图片中车辆特征提取模块2提取到车辆的车辆深度特征进行比对，并将计算结果发送给车辆违章检测模块4；车辆违章检测模块4收到车辆对比模块3的结果，并根据车辆检测模块1提取的相应车辆的坐标，用于与目标违章车辆的相同车辆定位，再根据具体定位判断违章是否为误判；As shown in Figure 1, a kind of automatic review system of illegal pictures based on vehicle retrieval of the present invention is made of vehicle detection module 1, vehicle feature extraction module 2, vehicle comparison module 3 and vehicle violation detection module 4, and the vehicle detection module 1. The vehicle comparison module 3 and the vehicle violation detection module 4 are respectively connected to the vehicle detection module 1 and the vehicle comparison module 3. The vehicle detection module 1 is responsible for vehicle detection of the violation pictures. The resolution of the violation pictures is 7296*5792. All the vehicles are detected; the vehicle feature extraction module 2 is used to extract the vehicle depth features from the vehicles detected by the vehicle detection module 1; the vehicle comparison module 3 calculates the matching degree of vehicle features, and is used to compare the vehicle of the target illegal vehicle judged by the system Feature is compared with the vehicle depth feature extracted by the vehicle feature extraction module 2 in the violation picture, and the calculation result is sent to the vehicle violation detection module 4; the vehicle violation detection module 4 receives the result of the vehicle comparison module 3, and according to the vehicle The coordinates of the corresponding vehicle extracted by the detection module 1 are used for the same vehicle location as the target illegal vehicle, and then judge whether the violation is a misjudgment according to the specific location;
所述车辆检测模块1的具体车辆定位采用实时目标检测网络YOLOv2实现的，该网络包括了24个卷积层与2个全连接层，提取出车辆的框为左上角端点坐标(X,Y)以及长W，宽H，令提取出的车辆框为C＝{carlisti|i＝1,2,...,n}，其中carlisti为第i辆的框的左上角端点坐标与长宽，n表示所检测出车的数量；The specific vehicle positioning of the vehicle detection module 1 is realized by the real-time target detection network YOLOv2, which includes 24 convolutional layers and 2 fully connected layers, and the frame of the extracted vehicle is the upper left corner endpoint coordinates (X, Y) And length W, width H, let the extracted vehicle frame be C={carlisti |i=1,2,...,n}, where carlisti is the coordinates and length and width of the upper left corner of the frame of the i-th vehicle , n represents the number of detected cars;
所述车辆特征提取模块2的车辆深度特征提取利用深度卷积神经网络实现，提取到的车辆特征为4096维的深度车辆特征，令提取出的车辆特征向量为F＝{feati|i＝1,2,...4096}，其中feati表示车辆特征向量中第i维的特征，为浮点数类型；The vehicle depth feature extraction of the vehicle feature extraction module 2 is realized by a deep convolutional neural network, and the extracted vehicle feature is a 4096-dimensional deep vehicle feature, so that the extracted vehicle feature vector is F={feati |i=1 ,2,...4096}, where feati represents the feature of the i-th dimension in the vehicle feature vector, which is a floating-point number type;
所述车辆对比模块3是根据公式(1)计算车辆之间的相似度Described vehicle comparison module 3 is to calculate the similarity between vehicles according to formula (1)
其中Q，P分别代表两辆车的特征向量，Q＝{qi|i＝1,2,...,4096}，P＝{pi|i＝1,2,...,4096}，qi表示特征向量Q中的第i个特征值，Pi表示特征向量P中的第i个特征值，K表示车辆特征维度，s表示Q,P两辆车的相似度函数；Where Q and P represent the feature vectors of the two vehicles respectively, Q={qi |i=1,2,...,4096}, P={pi |i=1,2,...,4096} , qi represents the i-th eigenvalue in the eigenvector Q, Pi represents the i-th eigenvalue in the eigenvector P, K represents the vehicle feature dimension, s represents the similarity function of Q and P two vehicles;
所述车辆违章检测模块4是根据车辆对比模块3计算的对比结果，若车辆之间的相似度s＞K0，K0为相似度阈值，则将此车辆信息(cn,carlistn)加入相似车辆集合所获得的违章目标车辆的相似车辆集合S，由于违章图片的特殊性，以图片中心点像素将图片分为四个区域(第一区域、第二区域、第三区域及第四区域)，目标违章车辆存在于第四区域，根据公式(2)获得车辆具体框的中心点坐标，确定相似车辆所在的位置，The vehicle violation detection module 4 is based on the comparison result calculated by the vehicle comparison module 3, if the similarity between vehicles s>K0 , K0 is the similarity threshold, then add the vehicle information (cn , carlistn ) to The similar vehicle set S of the illegal target vehicle obtained by the similar vehicle set, due to the particularity of the illegal picture, divides the picture into four areas (the first area, the second area, the third area and the fourth area) by the pixel of the center point of the picture. ), the target illegal vehicle exists in the fourth area, according to the formula (2) to obtain the coordinates of the center point of the specific frame of the vehicle, and determine the position of the similar vehicle,
其中，(Xc,Yc)是相似车辆中心点坐标，(X,Y)为车辆检测模块1检测出的相似车辆具体框上角端点坐标，W表示具体框的宽，H表示具体框的高，根据相似车辆中心点坐标判断车辆所在的具体区域，获得每个区域相似车辆集合S1＝{(cx,carlistx)|x＝1,2,...,n}，cx表示相似车辆的具体编号，令D＝max(S1)，获得D所在的相似车辆的具体框就是当前区域的目标违章车辆，判断第一区域、第二区域及第三区域中是否都存在目标违章车辆，如若都存在，则目标违章车辆确定违章，反之则为误检结果。Among them, (Xc , Yc ) are the coordinates of the center point of similar vehicles, (X, Y) are the coordinates of the upper corner endpoints of the specific frame of similar vehicles detected by the vehicle detection module 1, W represents the width of the specific frame, and H represents the width of the specific frame High, judge the specific area where the vehicle is located according to the coordinates of the center point of similar vehicles, and obtain the set of similar vehicles in each area S1 ={(cx ,carlistx )|x=1,2,...,n}, cx represents The specific number of the similar vehicle, set D=max(S1 ), the specific box of the similar vehicle where D is obtained is the target violation vehicle in the current area, and judge whether there is a target violation in the first area, the second area and the third area If all vehicles exist, the target violating vehicle is determined to be violating, otherwise it is a false detection result.
如图所示，基于本发明车辆检索的违章图片自动审核系统的车辆审核方法，包括如下步骤：As shown in the figure, the vehicle review method based on the automatic review system for illegal pictures retrieved by the vehicle of the present invention includes the following steps:
1)由于违章图片的特殊性，以图片中心点像素将图2将违章图片以图片中心点像素分为四个区域，分别为第一、第二、第三及第四区域，系统检测出的目标违章车辆存在于第四区域，令车辆检测模块(1)从违章图片的四个区域中提取出同张图片中所有车辆特征向量集合并检测出所有车辆与车辆特征提取模块2提取到所有车辆的车辆特征向量的映射集合为其中cj表示第j辆车的编号，n表示车辆数量，表示第j辆车的特征向量的集合，表示第j辆车的车辆特征集合中第k个车辆特征向量；1) Due to the particularity of the illegal picture, the illegal picture in Figure 2 is divided into four areas by the pixel of the center point of the picture, which are the first, second, third and fourth areas respectively. The target illegal vehicle exists in the fourth area, so that the vehicle detection module (1) extracts all vehicle feature vector sets in the same picture from the four areas of the illegal picture And detect all vehicles and vehicle feature extraction module 2 to extract the mapping set of vehicle feature vectors of all vehicles as Where cj represents the number of the jth vehicle, n represents the number of vehicles, Represents the set of feature vectors of the jth car, Represents the kth vehicle feature vector in the vehicle feature set of the jth vehicle;
2)由车辆特征提取模块2提取出第四区域目标违章车辆的特征向量Fm＝{feati|i＝1,2,...,4096}，i表示第i个车辆特征向量，并将目标违章车辆的特征向量信息传送给车辆对比模块3；2) The vehicle feature extraction module 2 extracts the feature vector Fm ={feati |i=1,2,...,4096} of the target illegal vehicle in the fourth area, where i represents the i-th vehicle feature vector, and The feature vector information of the target illegal vehicle is sent to the vehicle comparison module 3;
3)车辆对比模块3将提取到的目标违章车辆的车辆特征向量根据公式(1)与步骤1)中提取出的同张图片中所有车辆特征向量集合中的每个车辆特征向量的相似度，得到相似度集合若两辆车的相似度s＞K0，K0为车辆置信度阈值，则将此车辆信息(cn,carlistn)加入相似车辆集合，cn表示车辆具体编号，carlistn表示车辆所在位置的具体框，由此获得所有违章目标车辆的相似车辆；3) Vehicle comparison module 3 extracts the vehicle feature vector of the target illegal vehicle according to the set of all vehicle feature vectors in the same picture extracted in formula (1) and step 1) Each vehicle feature vector in The similarity, get the similarity set If the similarity of two vehicles s>K0 , K0 is the vehicle confidence threshold, then add the vehicle information (cn , carlistn ) to the set of similar vehicles, cn indicates the specific number of the vehicle, and carlistn indicates the location of the vehicle The specific box of , thereby obtaining similar vehicles of all violation target vehicles;
4)根据步骤3)中所获得的违章目标车辆的相似车辆集合S，获得相似车辆所在位置的具体框，根据公式(2)获得车辆具体框的中心点坐标，确定相似车辆所在的位置，4) According to the similar vehicle set S of the illegal target vehicle obtained in step 3), obtain the specific frame of the location of the similar vehicle, obtain the center point coordinates of the specific frame of the vehicle according to the formula (2), and determine the location of the similar vehicle,
其中，(Xc,Yc)是相似车辆中心点坐标，(X,Y)表示相似车辆具体框左顶点坐标，W表示具体框的宽，H表示具体框的高；Wherein, (Xc , Yc ) are the coordinates of the center point of similar vehicles, (X, Y) represent the coordinates of the left apex of the specific frame of similar vehicles, W represents the width of the specific frame, and H represents the height of the specific frame;
车辆违章检测模块4根据相似车辆中心点坐标判断车辆所在的具体区域，获得每个区域相似车辆集合S1＝{(cx,carlistx)|x＝1,2,...,n}，cx表示相似车辆的具体编号，令D＝max(S1)，获得D所在的相似车辆的具体框，就是当前区域的目标违章车辆，若一个区域中存在多辆目标违章车辆的相似车辆，则根据相似的大小选择相似度最大的相似车辆，判断为当前区域的目标违章车辆，将相似度较小的车辆排除出违章车辆判断，判断三个区域中是否都存在目标违章车辆如若都存在，则目标违章车辆确定违章，如图3所示；反之则为误检结果。The vehicle violation detection module 4 judges the specific area where the vehicle is located according to the coordinates of the center point of the similar vehicle, and obtains a set of similar vehicles in each area S1 ={(cx ,carlistx )|x=1,2,...,n}, cx represents the specific serial number of similar vehicles, let D=max(S1 ), the specific frame of the similar vehicle where D is obtained is the target violating vehicle in the current area, if there are multiple similar vehicles of the target violating vehicle in an area, Then select the similar vehicle with the largest similarity according to the size of the similarity, judge it as the target illegal vehicle in the current area, exclude the vehicle with a smaller similarity from the illegal vehicle judgment, and judge whether there are target illegal vehicles in the three areas. The target violating vehicle is determined to be violating the rules, as shown in Figure 3; otherwise, it is a false detection result.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910779521.4ACN110490150B (en) | 2019-08-22 | 2019-08-22 | A system and method for automatic review of illegal pictures based on vehicle retrieval |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910779521.4ACN110490150B (en) | 2019-08-22 | 2019-08-22 | A system and method for automatic review of illegal pictures based on vehicle retrieval |
| Publication Number | Publication Date |
|---|---|
| CN110490150Atrue CN110490150A (en) | 2019-11-22 |
| CN110490150B CN110490150B (en) | 2022-02-11 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910779521.4AActiveCN110490150B (en) | 2019-08-22 | 2019-08-22 | A system and method for automatic review of illegal pictures based on vehicle retrieval |
| Country | Link |
|---|---|
| CN (1) | CN110490150B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111429726A (en)* | 2020-04-13 | 2020-07-17 | 城云科技（中国）有限公司 | Monitoring video illegal parking vehicle detection and management method and corresponding system |
| CN111444901A (en)* | 2020-06-17 | 2020-07-24 | 爱保科技有限公司 | Processing method and device for distinguishing multiple vehicles from pictures in vehicle insurance case and electronic equipment |
| CN111767917A (en)* | 2019-12-24 | 2020-10-13 | 杭州海康威视数字技术股份有限公司 | Violation evidence picture detection method and device and electronic equipment |
| CN111768630A (en)* | 2019-12-24 | 2020-10-13 | 上海高德威智能交通系统有限公司 | A method, device and electronic device for detecting illegal and waste pictures |
| CN112101268A (en)* | 2020-09-23 | 2020-12-18 | 浙江浩腾电子科技股份有限公司 | A method of vehicle pressure line detection based on geometric projection |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130279742A1 (en)* | 2012-04-18 | 2013-10-24 | Ivs, Inc. | Automatic record detection device and method for stop arm violation event |
| US20150332097A1 (en)* | 2014-05-15 | 2015-11-19 | Xerox Corporation | Short-time stopping detection from red light camera videos |
| CN109948417A (en)* | 2018-12-31 | 2019-06-28 | 上海眼控科技股份有限公司 | A kind of vehicle based on deep learning drives in the wrong direction illegal automatic auditing method |
| CN109949578A (en)* | 2018-12-31 | 2019-06-28 | 上海眼控科技股份有限公司 | A kind of illegal automatic auditing method of vehicle crimping based on deep learning |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130279742A1 (en)* | 2012-04-18 | 2013-10-24 | Ivs, Inc. | Automatic record detection device and method for stop arm violation event |
| US20150332097A1 (en)* | 2014-05-15 | 2015-11-19 | Xerox Corporation | Short-time stopping detection from red light camera videos |
| CN109948417A (en)* | 2018-12-31 | 2019-06-28 | 上海眼控科技股份有限公司 | A kind of vehicle based on deep learning drives in the wrong direction illegal automatic auditing method |
| CN109949578A (en)* | 2018-12-31 | 2019-06-28 | 上海眼控科技股份有限公司 | A kind of illegal automatic auditing method of vehicle crimping based on deep learning |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111767917A (en)* | 2019-12-24 | 2020-10-13 | 杭州海康威视数字技术股份有限公司 | Violation evidence picture detection method and device and electronic equipment |
| CN111768630A (en)* | 2019-12-24 | 2020-10-13 | 上海高德威智能交通系统有限公司 | A method, device and electronic device for detecting illegal and waste pictures |
| CN111429726A (en)* | 2020-04-13 | 2020-07-17 | 城云科技（中国）有限公司 | Monitoring video illegal parking vehicle detection and management method and corresponding system |
| CN111429726B (en)* | 2020-04-13 | 2021-11-26 | 城云科技（中国）有限公司 | Monitoring video illegal parking vehicle detection and management method and corresponding system |
| CN111444901A (en)* | 2020-06-17 | 2020-07-24 | 爱保科技有限公司 | Processing method and device for distinguishing multiple vehicles from pictures in vehicle insurance case and electronic equipment |
| CN112101268A (en)* | 2020-09-23 | 2020-12-18 | 浙江浩腾电子科技股份有限公司 | A method of vehicle pressure line detection based on geometric projection |
| CN112101268B (en)* | 2020-09-23 | 2022-07-29 | 浙江浩腾电子科技股份有限公司 | Vehicle line pressing detection method based on geometric projection |
| Publication number | Publication date |
|---|---|
| CN110490150B (en) | 2022-02-11 |
| Publication | Publication Date | Title |
|---|---|---|
| CN110490150A (en) | A kind of automatic auditing system of picture violating the regulations and method based on vehicle retrieval | |
| CN107506763B (en) | An accurate positioning method of multi-scale license plate based on convolutional neural network | |
| CN107665603B (en) | Real-time detection method for judging parking space occupation | |
| CN105702048B (en) | Highway front truck illegal road occupation identifying system based on automobile data recorder and method | |
| CN104616502B (en) | Car license recognition and alignment system based on combination type bus or train route video network | |
| CN102855622B (en) | A kind of infrared remote sensing image sea ship detection method based on significance analysis | |
| CN109190444B (en) | An implementation method of a video-based toll lane vehicle feature recognition system | |
| CN103279756B (en) | Vehicle detection based on integrated classifier analyzes system and determination method thereof | |
| CN103077407B (en) | Car logo positioning and recognition method and car logo positioning and recognition system | |
| CN112651293B (en) | Video detection method for road illegal spreading event | |
| CN107491753A (en) | A kind of parking offense detection method based on background modeling | |
| CN102708691B (en) | False license plate identification method based on matching between license plate and automobile type | |
| CN102915433B (en) | Character combination-based license plate positioning and identifying method | |
| CN101770583B (en) | Template matching method based on global features of scene | |
| CN101630361A (en) | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles | |
| CN108052887A (en) | A kind of doubtful illegal land automatic recognition system and method for merging SLAM/GNSS information | |
| CN103927880A (en) | Vehicle license plate recognizing and matching method and device | |
| CN106446834A (en) | An image-based vehicle identification method and device | |
| CN113011331A (en) | Method and device for detecting whether motor vehicle gives way to pedestrians, electronic equipment and medium | |
| CN113158954B (en) | Automatic detection method for zebra crossing region based on AI technology in traffic offsite | |
| CN111582084A (en) | A method and system for detecting foreign objects on rails from a space-based perspective based on weakly supervised learning | |
| CN113822285A (en) | Vehicle illegal parking identification method for complex application scene | |
| CN105184291A (en) | Method and system for detecting multiple types of license plates | |
| CN116682268A (en) | Portable urban road vehicle violation inspection system and method based on machine vision | |
| CN107507140A (en) | The outdoor scene vehicle shadow disturbance restraining method of highway of feature based fusion |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |