CN114648748A - Motor vehicle illegal parking intelligent identification method and system based on deep learning - Google Patents
Motor vehicle illegal parking intelligent identification method and system based on deep learning Download PDFInfo
- Publication number
- CN114648748A CN114648748A CN202210559915.0A CN202210559915A CN114648748A CN 114648748 A CN114648748 A CN 114648748A CN 202210559915 A CN202210559915 A CN 202210559915A CN 114648748 A CN114648748 A CN 114648748A
- Authority
- CN
- China
- Prior art keywords
- motor vehicle
- vehicle
- frame
- target
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method and a system for intelligently identifying motor vehicle parking violation based on deep learning, which relate to the technical field of intelligent identification and aim to solve the technical problem that the motor vehicle parking violation identification based on computer vision is inaccurate in the prior art; performing transfer learning training based on a pre-training model of a target detection model YOLOV5 of the target detection field State-of-the-art; configuring algorithm parameters and illegal regions through a user configuration interface based on a web; sending each frame of image acquired by the camera to a target detection model to obtain a detection result of the motor vehicle target; sending the detection result of the motor vehicle target into an SORT tracking algorithm, and tracking the identified motor vehicle target; the invention is used for accurately and efficiently identifying the motor vehicle which is illegally parked.
Description
Technical Field
The invention relates to the technical field of intelligent recognition, in particular to a method and a system for intelligently recognizing motor vehicle parking violation based on deep learning.
Background
In recent years, computer vision technology has developed rapidly and has found effective use in numerous military and civilian applications. The target detection technology based on deep learning can automatically extract key characteristic information in a scene, and quickly and accurately identify and position a target. Consequently, the target detection can be discerned for the state of the motor vehicle in the wisdom garden, and the alignment is parked regional and is parked when long and judge to the realization need not personnel and keeps watch in real time, can accurately discern the motor vehicle of breaking to stop high-efficiently based on the sharp old camera in garden.
As an important application scenario in the smart park, the identification of motor vehicle parking violation based on computer vision is widely used at present. However, the problems of inaccurate identification, high false alarm rate, high missing report rate and the like at present seriously hinder user experience, so that an accurate and efficient motor vehicle illegal parking identification is urgently needed.
Disclosure of Invention
The invention aims to provide a motor vehicle illegal parking intelligent identification method and system based on deep learning, and aims to solve the technical problems that the motor vehicle illegal parking identification based on computer vision is inaccurate and the false alarm rate and the missing report rate are high in the prior art.
The technical effects that can be produced by the preferred technical scheme in the technical schemes provided by the invention are described in detail in the following.
In order to achieve the purpose, the invention provides the following technical scheme:
the invention provides a motor vehicle illegal parking intelligent identification method and system based on deep learning, which comprises the following steps:
step 100, collecting a data image for training, and manually labeling all motor vehicles in the data image to obtain a data set for training;
step 200, performing transfer learning training on a pre-training model of a target detection model YOLOV5 based on a target detection field State-of-the-art by using the data set;
step 300, configuring algorithm parameters and illegal parking areas through a user configuration interface based on a web;
step 400, acquiring video data of a camera, and sending each frame of image acquired by the camera into the target detection model to obtain a motor vehicle target detection result;
step 500, sending the motor vehicle target detection result into an SORT tracking algorithm, and tracking the identified motor vehicle target;
and step 600, identifying and judging whether the tracked motor vehicle target is located in the illegal parking area and judging whether the motor vehicle target is illegal.
Preferably, after the transfer learning training is carried out on the pre-training model of the target detection model Yolov5 based on the target detection field State-of-the-art, the optimization operations of operator fusion, Kernal funcition optimization and weight quantification are carried out on the model by using TensorRT.
Preferably, the judging whether the vehicle is in the parking violation area comprises the following steps:
601, acquiring a left lower foot point, a right lower foot point and a detection frame center point coordinate of a certain motor vehicle detection frame in a current image frame;
step 602, if the left lower foot point and the right lower foot point of the motor vehicle detection frame are both in the parking violation area, judging that the motor vehicle is parked illegally;
step 603, if only one side foot point of the motor vehicle detection frame is located in the illegal parking area, performing step 604;
step 604, judging whether the vehicle center point is located in an illegal parking area, if so, judging that the motor vehicle is illegal to park, otherwise, performing the next step;
step 605, continuing to acquire other detection targets in the current image frame and performing the operation of judging whether the vehicle is illegal to stop.
The application also comprises a method for judging whether the motor vehicle is in the illegal parking area to move slowly, which comprises the following steps:
step a01, calculating the pixel distance between the center coordinates of the motor vehicle in the previous frame image and the center coordinates of the motor vehicle in the current frame image, and if the pixel distance exceeds a threshold value, determining that the vehicle moves;
step a02, if the vehicle is judged to be in a moving state, clearing the illegal parking time length in the vehicle information and processing other detection target vehicles;
step a03, if the vehicle is not moving, acquiring the current time and calculating the vehicle stop accumulated time, comparing the vehicle stop accumulated time with a set violation time threshold, if the vehicle stop accumulated time exceeds the violation time threshold, determining that the vehicle is violating, if the vehicle stop accumulated time does not exceed the violation time, ending the processing of the current target vehicle, and continuing to process other detection targets.
Preferably, the threshold value adopts a dynamic form threshold value of d/p to judge that the vehicle moves slowly, wherein the letter d represents the diagonal distance of the detection box, and the letter p is a threshold value scale factor.
Preferably, if a certain vehicle is tracked by a previous frame of image in the tracking information and the vehicle is not tracked in a current frame, the system sets a maximum vanishing frame number, the motor vehicle is not judged to be lost before the maximum vanishing frame number is not reached, then, a Kalman filtering in an SORT tracking algorithm is used for predicting the motor vehicle target frame of the current frame based on the motor vehicle target frame position in the previous frame, and the predicted result is used as the current frame target frame;
if the next frame of image detects that the position of the motor vehicle target frame is matched with the current frame of target frame, judging that the vehicle disappears caused by error of the detection algorithm;
if the maximum number of lost frames is reached, directly regarding that the motor vehicle target is lost, and deleting the target tracking information by the system;
and if the motor vehicle target reappears in the time period of not reaching the maximum vanishing frame number, judging that the motor vehicle target is shielded for a short time.
Preferably, the maximum number of lost frames is recommended to be set to 250-500 frames, i.e. 10-20 seconds.
The system of the application comprises a detection model module, an image processing module and a violation judging module, wherein:
the detection model module is used for carrying out target detection on the camera video data acquired by the image processing module and obtaining a detection frame and detection frame information of the vehicle;
the image processing module is used for acquiring camera video data, configuring the position of an forbidden zone frame, matching a unique ID (identity) with the information of a vehicle detection frame and a detection frame obtained by the detection model module and then transmitting the information to the illegal parking judgment module;
and the illegal parking judgment module outputs tracker information after receiving the vehicle detection frame and the detection frame information matched with the unique ID, searches for each tracker information, judges whether the vehicle is positioned in an forbidden zone according to the position of the forbidden zone frame, updates the tracker information again, and judges whether the vehicle should be alarmed according to the updated tracker information.
Preferably, the tracker information includes: vehicle ID, vehicle detection box, whether the vehicle entered the parking violation area, when the vehicle entered the parking violation area, whether the vehicle has been alerted.
Preferably, the data set can be enhanced in a translation, flip, and zoom manner, i.e., the amount of training data is increased.
The invention applies the target detection technology of deep learning to the illegal parking recognition of the motor vehicle, and the method can accurately recognize the motor vehicle target from the image, track the motor vehicle target, accurately and efficiently recognize the illegal parking motor vehicle and generate an alarm through a series of logic judgment of illegal parking of the motor vehicle, thereby realizing the intelligent supervision of the illegal parking of the motor vehicle. Based on the intelligent illegal parking recognition system provided by the invention, the staff can obtain the warning message of illegal parking of the motor vehicle in the park in real time without regular and fixed-point patrol, thereby greatly improving the efficiency of the staff and saving a large amount of manpower and material resources;
a detection model module: the invention adopts a State-of-the-art detection model YOLOV5 to carry out customized training, thereby ensuring the detection accuracy of the motor vehicle. The method uses the neural network model to perform throughput performance optimization through the functions of operator fusion, Kernal funcition optimization, weight quantization processing and the like of an Nvidia TensorRT network model compression tool so as to ensure the performance optimization under the limited budget;
an image processing module: the method comprises the steps of tracking a detected motor vehicle based on a tracking algorithm (SORT), maintaining an individual structure body of the motor vehicle, judging the state of the motor vehicle, and setting a tracking failure threshold value to avoid false alarm after the motor vehicle is shielded;
a violation judgment module: the method comprises the steps of firstly judging whether a tracked motor vehicle is in a forbidden zone, providing various judging modes for slowly moving the motor vehicle, and judging the slowly moving motor vehicle so as to avoid the problem of false alarm of the slowly moving motor vehicle in the traditional method;
the invention can conveniently configure the illegal parking recognition algorithm from the web page, and the alarm message can be displayed in the web page in real time for the staff to check, thereby forming an intelligent motor vehicle illegal parking recognition system together with the core detection recognition algorithm.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of an embodiment of a method and a system for intelligent recognition of vehicle parking violation based on deep learning according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. It should be apparent that the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the examples given herein without any inventive step, are within the scope of the present invention.
The following describes in detail a specific embodiment of the present invention with reference to the drawings. In the drawings, the same reference numerals indicate the same or corresponding features. The figures are only schematic and are not necessarily drawn to scale.
The flow chart of an embodiment of the intelligent motor vehicle parking violation identification method and system based on deep learning is shown in figure 1,
detection model module
Firstly, data used for training are collected based on an actual scene, and after the data are obtained, a data set used for training is obtained by manually marking all motor vehicles in a data image.
In order to better ensure the training effect, the invention can greatly increase the training data volume by adopting data enhancement modes such as translation, turnover, scaling and the like for the training data set.
And then carrying out transfer learning training on a pre-training model of a target detection model YOLOV5 based on the target detection field State-of-the-art through a training data set so as to ensure the accuracy of the model.
After the trained model is obtained, the traditional method generally directly deploys and uses the trained original model in combination with a service program, and the mode not only has low inference speed, but also extremely occupies hardware resources.
In order to ensure the real-time performance of target detection in the service, the original model uses TensorRT to perform operations of operator fusion, Kernal funcition optimization, weight quantization and the like on the model to optimize the model inference throughput, and forward inference is performed to accelerate the inference.
Image processing module
Firstly, the intelligent identification system configures algorithm parameters through a user configuration interface based on a web, a web page displays a picture of a camera to be configured in real time, a user can draw an illegal parking area in the picture through a mouse, and after the user finishes drawing, the configuration of the algorithm is sent to an algorithm server through a network.
After the algorithm receives the configuration, the real-time data of the configuration camera is pulled according to the appointed video stream address, and each frame of obtained image is sent to a trained target detection model to obtain the target detection result of the motor vehicle.
And then, sending the target detection result into an SORT tracking algorithm to track the identified motor vehicle target.
The tracking has the advantages that the same target in the time sequence is matched and is endowed with the same ID, and the problem of continuous alarm of the same vehicle is avoided.
Illegal parking judgment module
Traversing all tracked targets according to the tracker tracking information of the image processing module, and initializing the information if the tracked targets are newly tracked vehicles;
if the vehicle is tracked by the previous system and the vehicle is tracked by the current frame, firstly judging whether the vehicle is positioned in the illegal parking area, and judging whether the vehicle is positioned in the illegal parking area by adopting the coordinates of the left lower pin point, the right lower pin point and the central point of the detection frame of the motor vehicle.
The specific method comprises the following steps: firstly, acquiring coordinates of a pin point at the left lower side and the right lower side of a detection frame bounding box and a center point of the detection frame;
if the left lower foot point and the right lower foot point are both in the parking violation area, judging that the motor vehicle is parked illegally;
if only one side foot point is located in the illegal parking area, judging whether the vehicle center point is located in the illegal parking area, and if so, judging that the motor vehicle is illegal to park; the rest are considered as being free of violations.
And if the motor vehicle is not positioned in the illegal parking area, finishing the processing of the current target vehicle, continuously processing other detection targets, finishing the processing of the current frame after all the targets are traversed, and continuously acquiring the next frame.
If the motor vehicle is located in the illegal parking area, the invention adopts a method for judging whether the motor vehicle moves slowly, so as to avoid the problem that the vehicle is always in the illegal parking area in the process of slowly driving in a large illegal parking area due to traffic jam or other reasons, and is mistakenly considered as illegal parking to give an alarm.
The specific method comprises the following steps:
(1) calculating the central coordinates of the motor vehicle in the previous frame of image and the central coordinates of the motor vehicle in the current frame of image to calculate the pixel distance, and if the distance exceeds a certain threshold value, judging that the vehicle moves;
(2) the sizes of the boxes detected by the motor vehicles with different distances from the camera in the image have great difference due to different distances from the vehicle to the camera; for example, the actual road surface distance corresponding to the distance of 10 pixels far away is different from the actual road surface distance corresponding to the distance of 10 pixels near.
Therefore, if the method in step (1) uses a single threshold as the determination condition, it may be determined that the creep criterion is different for different vehicles at different distances.
The invention provides a method for judging whether a vehicle moves slowly by using a dynamic threshold = d/p form;
wherein d is the diagonal distance of the detection frame and p is a threshold value scale factor;
and p is a fixed value in the judgment process and can be adjusted by an algorithm worker.
The dynamic threshold is set up, so that the corresponding distance threshold is larger when the detection frame is larger, and the corresponding distance threshold is smaller when the detection frame is smaller, so that the movement judgment of the vehicles at different distances from the camera is more accurate.
If the vehicle moves, clearing the illegal stopping time in the vehicle information for a long time, and continuously processing other detection targets;
if the vehicle is not moving, acquiring the current time and calculating the vehicle stopping accumulated time, comparing the vehicle stopping accumulated time with the illegal stopping time threshold, if the vehicle stopping accumulated time exceeds the illegal stopping time threshold, judging that the vehicle is illegal, sending vehicle information to a web end through a network, and displaying alarm information on a page after the web end receives the alarm information; if the time length does not exceed the illegal parking time length, the current target vehicle is processed, and other detection targets are continuously processed.
If a certain vehicle is tracked in the previous frame of the tracking information and the current frame is not tracked, the following 3 situations occur:
1. detecting that the target is not detected due to algorithm errors;
2. the target exits the range of the picture;
3. and when the target is blocked by other objects, the target cannot be detected by a detection algorithm.
In view of the above situation, the specific method of the present invention is: and setting a maximum number of vanishing frames by combining the three possible situations, and not judging that the target is lost before the maximum number of the vanishing frames is not reached. And predicting the target frame of the current frame by using Kalman filtering in the SORT tracking algorithm based on the position of the target frame of the previous frame, wherein the predicted result is used as the target frame of the current frame, but the tracking state is still not tracked.
For the first situation, if the target is not detected due to the error of the detection algorithm, when the target frame is detected in the next frame, the target frame obtained by the tracking algorithm is matched with the target frame of the current frame, so that the problem of repeated alarm caused by the instability of the detection algorithm is avoided;
aiming at the second condition, when the maximum number of vanishing frames is reached, the algorithm regards that the current vehicle target disappears, and the target tracking information is deleted;
for the third situation, if the target is shielded for a short time, the tracking algorithm after the target reappears can still be identified as the same vehicle based on the matching of the previous target frame and the shielded target frame, so that the problem of repeated alarming is avoided.
The maximum number of lost frames is suggested to be set to 250-500 frames, i.e., 10-20 seconds.
The specific detection model module, the image processing module and the illegal parking judging module are communicated through the following contents:
the detection model module generates a model file after training and accelerating;
the image processing module firstly acquires web configuration through network communication; the web configuration specifically comprises: camera rtsp stream address and forbidden zone frame position;
after configuration is completed, the image processing module loads a model file of the detection model module to perform target detection on the video data of the camera and obtain a detection frame and detection frame information of the vehicle;
sending the detection frames into an SORT tracking algorithm, matching a unique ID for each detection frame, and then sending the detection frames into an illegal parking judgment module to obtain tracker information, wherein the tracker information comprises a vehicle ID, a vehicle detection frame, whether the vehicle enters an illegal parking area, the time when the vehicle enters the illegal parking area, and whether the vehicle is warned;
the illegal parking judgment module traverses each tracker information, judges whether the vehicle is positioned in an forbidden zone according to the position of the forbidden zone frame, updates the tracker information and judges whether the vehicle needs to give an alarm according to the tracker information.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (10)
1. A motor vehicle illegal parking intelligent identification method and system based on deep learning are characterized in that: the method comprises the following steps:
step 100, collecting a data image for training, and manually labeling all motor vehicles in the data image to obtain a data set for training;
step 200, performing transfer learning training on a pre-training model of a target detection model YOLOV5 based on a target detection field State-of-the-art by using the data set;
step 300, configuring algorithm parameters and illegal parking areas through a user configuration interface based on a web;
step 400, acquiring video data of a camera, and sending each frame of image acquired by the camera into the target detection model to obtain a motor vehicle target detection result;
step 500, sending the motor vehicle target detection result into an SORT tracking algorithm, and tracking the identified motor vehicle target;
and step 600, identifying and judging whether the tracked motor vehicle target is located in the illegal parking area and judging whether the motor vehicle target is illegal.
2. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 1, wherein: after the pre-training model of the target detection model YOLOV5 based on the target detection field State-of-the-art is subjected to transfer learning training, the model is subjected to optimization operations of operator fusion, Kernal funcition optimization and weight quantization by using TensorRT.
3. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 1, wherein: determining whether the vehicle is in the parking violation area comprises the steps of:
601, acquiring a left lower foot point, a right lower foot point and a detection frame center point coordinate of a certain motor vehicle detection frame in a current image frame;
step 602, if the left lower foot point and the right lower foot point of the motor vehicle detection frame are both in the parking violation area, judging that the motor vehicle is parked illegally;
step 603, if only one side foot point of the motor vehicle detection frame is located in the illegal parking area, performing step 604;
step 604, judging whether the vehicle center point is located in an illegal parking area, if so, judging that the motor vehicle is illegal to park, otherwise, performing the next step;
step 605, continuing to acquire other detection targets in the current image frame and performing the operation of judging whether the vehicle is illegal to stop.
4. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 3, wherein: the method for judging whether the motor vehicle is in the parking violation area to move slowly comprises the following steps:
step a01, calculating the pixel distance between the center coordinates of the motor vehicle in the previous frame image and the center coordinates of the motor vehicle in the current frame image, and if the pixel distance exceeds a threshold value, determining that the vehicle moves;
step a02, if the vehicle is judged to be in a moving state, clearing the illegal parking time length in the vehicle information and processing other detection target vehicles;
step a03, if the vehicle is not moving, acquiring the current time and calculating the vehicle stop accumulated time, comparing the vehicle stop accumulated time with a set violation time threshold, if the vehicle stop accumulated time exceeds the violation time threshold, determining that the vehicle is violating, if the vehicle stop accumulated time does not exceed the violation time, ending the processing of the current target vehicle, and continuing to process other detection targets.
5. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 4, wherein: the threshold value adopts a dynamic form threshold value of d/p to judge that the vehicle slowly moves, wherein the letter d represents the diagonal distance of the detection frame, and the letter p is a threshold value scale factor.
6. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 4, wherein: if a certain vehicle is tracked by a previous frame image in the tracking information and the vehicle is not tracked in the current frame, setting a maximum vanishing frame number by the system, not judging that the motor vehicle is lost before the maximum vanishing frame number is not reached, then predicting the motor vehicle target frame of the current frame by using Kalman filtering in an SORT tracking algorithm based on the motor vehicle target frame position in the previous frame, and taking the predicted result as the current frame target frame;
if the next frame of image detects that the position of the motor vehicle target frame is matched with the current frame of target frame, judging that the vehicle disappears caused by the error of the detection algorithm;
if the maximum number of lost frames is reached, directly regarding that the motor vehicle target is lost, and deleting the target tracking information by the system;
and if the motor vehicle target reappears in the time period of not reaching the maximum vanishing frame number, judging that the motor vehicle target is shielded for a short time.
7. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 6, wherein: the time for playing the 250-500 frames of images is 10 seconds to 20 seconds.
8. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 1, wherein: the system comprises a detection model module, an image processing module and an illegal parking judgment module,
the detection model module is used for carrying out target detection on the camera video data acquired by the image processing module and obtaining a detection frame and detection frame information of the vehicle;
the image processing module is used for acquiring camera video data, configuring the position of an forbidden zone frame, matching a unique ID (identity) with the information of a vehicle detection frame and a detection frame obtained by the detection model module and then transmitting the information to the illegal parking judgment module;
and the illegal parking judgment module outputs tracker information after receiving the vehicle detection frame and the detection frame information matched with the unique ID, searches for each tracker information, judges whether the vehicle is positioned in an forbidden zone according to the position of the forbidden zone frame, updates the tracker information again, and judges whether the vehicle should be alarmed according to the updated tracker information.
9. The intelligent motor vehicle parking violation identification method and system based on deep learning as claimed in claim 8, wherein: the tracker information includes: vehicle ID, vehicle detection box, whether the vehicle entered the parking violation area, when the vehicle entered the parking violation area, whether the vehicle has been alerted.
10. The intelligent motor vehicle parking violation identification method and system based on deep learning of claim 1, wherein: the data set may be translated, flipped, scaled data enhancements-increasing the amount of training data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210559915.0A CN114648748A (en) | 2022-05-23 | 2022-05-23 | Motor vehicle illegal parking intelligent identification method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210559915.0A CN114648748A (en) | 2022-05-23 | 2022-05-23 | Motor vehicle illegal parking intelligent identification method and system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114648748A true CN114648748A (en) | 2022-06-21 |
Family
ID=81997600
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210559915.0A Pending CN114648748A (en) | 2022-05-23 | 2022-05-23 | Motor vehicle illegal parking intelligent identification method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114648748A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049948A (en) * | 2022-08-15 | 2022-09-13 | 深圳市万物云科技有限公司 | Unmanned aerial vehicle inspection method and device based on neural network model and related equipment |
| CN115082903A (en) * | 2022-08-24 | 2022-09-20 | 深圳市万物云科技有限公司 | Non-motor vehicle illegal parking identification method and device, computer equipment and storage medium |
| CN115994781A (en) * | 2023-03-23 | 2023-04-21 | 南通全织道家纺科技有限公司 | Accurate passenger acquisition system and method based on big data |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109615868A (en) * | 2018-12-20 | 2019-04-12 | 北京以萨技术股份有限公司 | A kind of video frequency vehicle based on deep learning is separated to stop detection method |
| CN110136449A (en) * | 2019-06-17 | 2019-08-16 | 珠海华园信息技术有限公司 | Traffic video frequency vehicle based on deep learning disobeys the method for stopping automatic identification candid photograph |
| CN110472496A (en) * | 2019-07-08 | 2019-11-19 | 长安大学 | A kind of traffic video intelligent analysis method based on object detecting and tracking |
| 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 |
| CN111539359A (en) * | 2020-04-28 | 2020-08-14 | 浙江工商大学 | Illegal parking detection method based on deep learning |
| CN113593250A (en) * | 2021-07-12 | 2021-11-02 | 浙江工贸职业技术学院 | Illegal parking detection system based on visual identification |
| CN113822285A (en) * | 2021-09-29 | 2021-12-21 | 重庆市云迈科技有限公司 | Vehicle illegal parking identification method for complex application scene |
-
2022
- 2022-05-23 CN CN202210559915.0A patent/CN114648748A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109615868A (en) * | 2018-12-20 | 2019-04-12 | 北京以萨技术股份有限公司 | A kind of video frequency vehicle based on deep learning is separated to stop detection method |
| CN110136449A (en) * | 2019-06-17 | 2019-08-16 | 珠海华园信息技术有限公司 | Traffic video frequency vehicle based on deep learning disobeys the method for stopping automatic identification candid photograph |
| CN110472496A (en) * | 2019-07-08 | 2019-11-19 | 长安大学 | A kind of traffic video intelligent analysis method based on object detecting and tracking |
| 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 |
| CN111539359A (en) * | 2020-04-28 | 2020-08-14 | 浙江工商大学 | Illegal parking detection method based on deep learning |
| CN113593250A (en) * | 2021-07-12 | 2021-11-02 | 浙江工贸职业技术学院 | Illegal parking detection system based on visual identification |
| CN113822285A (en) * | 2021-09-29 | 2021-12-21 | 重庆市云迈科技有限公司 | Vehicle illegal parking identification method for complex application scene |
Non-Patent Citations (1)
| Title |
|---|
| 姚巍巍: "基于模型剪枝和半精度加速改进YOLOv3-tiny算法的实时司机违章行为检测", 《计算机系统应用》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049948A (en) * | 2022-08-15 | 2022-09-13 | 深圳市万物云科技有限公司 | Unmanned aerial vehicle inspection method and device based on neural network model and related equipment |
| CN115049948B (en) * | 2022-08-15 | 2022-11-22 | 深圳市万物云科技有限公司 | Unmanned aerial vehicle inspection method and device based on neural network model and related equipment |
| CN115082903A (en) * | 2022-08-24 | 2022-09-20 | 深圳市万物云科技有限公司 | Non-motor vehicle illegal parking identification method and device, computer equipment and storage medium |
| CN115994781A (en) * | 2023-03-23 | 2023-04-21 | 南通全织道家纺科技有限公司 | Accurate passenger acquisition system and method based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108062349B (en) | Video monitoring method and system based on video structured data and deep learning | |
| CN102073851B (en) | Method and system for automatically identifying urban traffic accident | |
| CN114648748A (en) | Motor vehicle illegal parking intelligent identification method and system based on deep learning | |
| KR101095528B1 (en) | An outomatic sensing system for traffic accident and method thereof | |
| US11380105B2 (en) | Identification and classification of traffic conflicts | |
| WO2022227766A1 (en) | Traffic anomaly detection method and apparatus | |
| CN111564015B (en) | Method and device for monitoring perimeter intrusion of rail transit | |
| Giannakeris et al. | Speed estimation and abnormality detection from surveillance cameras | |
| CN104200466A (en) | Early warning method and camera | |
| CN104966304A (en) | Kalman filtering and nonparametric background model-based multi-target detection tracking method | |
| Liu et al. | Vision-based real-time lane marking detection and tracking | |
| CN111325048B (en) | Personnel gathering detection method and device | |
| CN104616502A (en) | License plate identification and positioning system based on combined type vehicle-road video network | |
| CN113822285A (en) | Vehicle illegal parking identification method for complex application scene | |
| CN112101223A (en) | Detection method, device, equipment and computer storage medium | |
| CN112560546B (en) | Method and device for detecting throwing behavior and storage medium | |
| CN113469115A (en) | Method and apparatus for outputting information | |
| CN113869275A (en) | Vehicle object detection system that throws based on remove edge calculation | |
| CN116311166A (en) | Traffic obstacle recognition method and device and electronic equipment | |
| CN113538968B (en) | Method and apparatus for outputting information | |
| CN111291722A (en) | Vehicle weight recognition system based on V2I technology | |
| Płaczek | A real time vehicle detection algorithm for vision-based sensors | |
| Kamijo et al. | Development and evaluation of real-time video surveillance system on highway based on semantic hierarchy and decision surface | |
| CN114708544A (en) | Intelligent violation monitoring helmet based on edge calculation and monitoring method thereof | |
| CN114445787A (en) | Non-motor vehicle weight recognition method and related equipment |
Legal Events
| 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220621 |