CN111797803A - Road guardrail abnormity detection method based on artificial intelligence and image processing - Google Patents
Road guardrail abnormity detection method based on artificial intelligence and image processing Download PDFInfo
- Publication number
- CN111797803A CN111797803A CN202010677433.6A CN202010677433A CN111797803A CN 111797803 A CN111797803 A CN 111797803A CN 202010677433 A CN202010677433 A CN 202010677433A CN 111797803 A CN111797803 A CN 111797803A
- Authority
- CN
- China
- Prior art keywords
- guardrail
- isolation
- curve
- image
- points
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- 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
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
Abstract
The invention provides a road guardrail abnormity detection method based on artificial intelligence and image processing, which comprises the following steps: the method comprises the steps of performing image splicing on collected road isolation guardrail images and projecting the images into a CIM of a road which is constructed in advance; performing semantic segmentation operation on the acquired image based on a semantic perception network to obtain a guardrail semantic segmentation image; operating the semantic guardrail segmentation graph based on the pixel classification network, acquiring pixel points which are judged as isolation guardrails in each row of pixels to obtain a guardrail scatter diagram, and fitting the scatter diagram to obtain an isolation guardrail curve; and after image splicing is carried out on the images including the curves of the isolation guardrails, the images are projected into the CIM, the CIM is subjected to visual processing by utilizing a Web GIS technology, and when the isolation guardrails are detected to be abnormal, warning information is displayed on a CIM visual interface. The method can monitor the isolation guardrail in real time, and related personnel can take measures in time when the isolation guardrail has abnormal conditions, so that road congestion is avoided.
Description
Technical Field
The invention relates to the field of artificial intelligence and intelligent traffic, in particular to a road guardrail abnormity detection method based on artificial intelligence and image processing.
Background
The isolation net guardrail in the middle of the road can effectively prevent the problems that non-motor vehicles randomly change lanes, pedestrians randomly pass through the isolation net guardrail and the like, so that traffic is more orderly. However, many isolation guardrails are moved privately or move and deform under the influence of external force, and occupy lanes in serious situations, so that traffic jam is caused, the traffic of vehicles is influenced, and traffic safety hazards also exist while inconvenience is brought to life of people.
The existing method for detecting the abnormity of the isolation guardrail is to install a three-axis accelerometer on the guardrail and analyze the change of the inclination angle of the guardrail detected by the three-axis accelerometer, thereby judging whether the isolation guardrail has abnormal conditions or not.
Disclosure of Invention
Aiming at the problems, the invention provides a road guardrail abnormity detection method based on artificial intelligence and image processing, which comprises the following steps:
the method comprises the following steps that firstly, the acquired road isolation guardrail images are subjected to distortion removal processing and then are spliced, and then the images are projected into a CIM of a road which is constructed in advance;
performing semantic segmentation operation on the acquired image based on a semantic perception network, and perceiving and segmenting the isolation guardrail to obtain a guardrail semantic segmentation image;
selecting semantic segmentation graphs containing roads, isolation guardrails and other classes as a training data set; marking each row of isolation guardrails as a curve with the width of 1 pixel; training a pixel classification network by using a cross entropy loss function;
operating the semantic guardrail segmentation graph based on the pixel classification network, acquiring pixel points which are judged as isolation guardrails in each row of pixels to obtain a guardrail scatter diagram, and fitting the scatter diagram to obtain an isolation guardrail curve; the pixel classification network comprises a classification encoder and a full-connection layer, wherein the input of the classification encoder is a guardrail semantic segmentation graph, and the guardrail semantic segmentation graph is sent into the full-connection layer to classify pixels after characteristics are extracted, wherein each row of pixels in the image respectively corresponds to one full-connection layer, and one full-connection layer is used for judging whether pixel points in one row of pixels are isolation guardrails or not;
fourthly, the images including the curves of the isolation guardrails are subjected to image splicing and then projected into the CIM, the CIM is subjected to visualization processing by utilizing a Web GIS technology, and when the isolation guardrails are detected to be abnormal, warning information is displayed on a CIM visualization interface; wherein, the unusual detection's of isolation barrier concrete mode does: and setting a distance threshold value, and judging that the isolation guardrail is abnormal when the distance between the deviation scatter outside the curve of the isolation guardrail and the curve is greater than the threshold value.
The semantic perception network comprises a semantic segmentation encoder and a semantic segmentation decoder, and the image containing the isolation guardrail in the road is selected as a training data set; labeling the data set, wherein the labeled road is 1, the isolation guardrail is 2, and the other classes are 3; training the network using a cross entropy loss function; the method comprises the steps that a feature map is obtained through a semantic segmentation encoder after a collected road isolation guardrail image is subjected to distortion removal processing, and a semantic segmentation decoder processes the feature map to obtain a guardrail semantic segmentation map.
The fitting of the scatter diagram comprises the following specific steps: randomly selecting N points in the scatter diagram to fit into a curve, calculating the distance from all the points to the curve, setting a threshold value, judging the points with the distance less than the threshold value as belonging to the curve, and recording the number of the points belonging to the curve; then randomly selecting N points, fitting a new curve again, and calculating the number of the points belonging to the curve according to the steps; finally, the curve with the largest number of points belonging to the curve is selected as the curve of the isolation barrier.
The image splicing method comprises the following steps: and after extracting the characteristic points, matching the characteristic points, screening correct matching point pairs to obtain a projection matrix, and performing image deformation and image fusion operation after performing registration operation according to the projection matrix.
The invention has the beneficial effects that:
1. the isolation guardrail is monitored by adopting the neural network technology, physical equipment does not need to be installed on the isolation guardrail, investment is saved, a large amount of human resources are not needed to maintain normal operation of the physical equipment in the follow-up process, the isolation guardrail is identified and extracted by adopting the neural network, the influence of other obstacles on a road is small, the identification accuracy is high, the extraction speed is high, and the operation of the neural network is not influenced when the isolation guardrail is damaged.
2. Sensing the isolation guardrail by adopting a semantic segmentation model, wherein the guardrail in the obtained result image is in an irregular strip shape, has poor continuity and serious adhesion condition, and is not convenient for accurate identification; when accurate curve information is required to be obtained, post-processing is usually performed in modes such as clustering and the like, but a guardrail curve is extracted in an image processing mode, a threshold value set by manual intervention is often required, generalization capability is poor, and the obtained curve is not accurate enough; according to the invention, a pixel classification network is added behind the semantic segmentation network, so that scattered point information of the isolation guardrail can be more accurately obtained, then more accurate curve information can be obtained by fitting scattered points, whether the isolation guardrail has abnormal conditions can be more effectively and accurately obtained after the processing of the pixel classification network, and the error probability of a network monitoring result is reduced.
3. The method comprises the steps of projecting a panoramic image of a road obtained after splicing and an isolation guardrail curve graph into a CIM to be combined, performing visual processing by utilizing a Web GIS technology, obtaining state information of the isolation guardrail more visually, realizing real-time monitoring of the isolation guardrail, and warning on a visual interface when the isolation guardrail moves or deforms, so that related personnel can take related measures in time, and road congestion caused by abnormal conditions of the isolation guardrail is avoided.
Drawings
Fig. 1 is an overall framework diagram of the method.
Detailed Description
The invention is described in detail below with reference to the examples and the accompanying drawings, in which reference is made to fig. 1.
Example (b):
a road guardrail abnormity detection method based on artificial intelligence and image processing is disclosed, the overall frame diagram of which is shown in figure 1, and the method comprises the following steps:
the road isolation guardrail image that gathers to fisheye camera carries out the image mosaic and obtains the panoramic picture of road after the distortion removal processing, later the projection to the CIM of the road of building in advance, specifically:
the method for correcting the distortion of the image mainly comprises a fixed inner diameter method, a fixed outer diameter method, a radial expansion method, an improved version double longitude method, a longitude and latitude mapping method and the like, and an implementer can select which method to use for distortion removal.
The image splicing method comprises the following steps: and after extracting the characteristic points, matching the characteristic points, screening correct matching point pairs to obtain a projection matrix, and performing image deformation and image fusion operation after performing registration operation according to the projection matrix.
The embodiment uses ORB algorithm to extract and describe the features, the speed of the ORB algorithm is obviously improved compared with SIFT algorithm and SURF algorithm, the ORB adopts FAST algorithm to detect and extract the feature points, and BRIEF algorithm is adopted to calculate a descriptor of the feature points.
The core idea of the FAST algorithm is to compare a point with its surrounding points, and consider it a feature point if the point is not the same as most of its surrounding points. The specific situation is as follows:
selecting a pixel point P from the image, wherein the density (gray value) of the pixel point P is set as Ip;
an appropriate threshold value t is set, and when the absolute value of the difference between the gray values of the point P and any point around the point P is larger than t, the 2 points are considered to be different.
Considering 16 pixels around the pixel point P, if there are n continuous points in the 16 points and the density difference between the point P and the density of the n continuous points is large, the point P is considered as a characteristic point, namely
Ix-Ip>t
In the formula IxRepresenting the density, I, of the xth pixelpThe density of the point P, t is a set threshold.
In order to reduce the amount of calculation and increase the calculation speed, in the embodiment, n is 4, after 16 pixel points around a point P are selected clockwise or counterclockwise, 4 pixels located in the directions of right above, right below, right left and right of the point P are selected from the 16 pixel points for judgment, and if the density difference between at least 3 of the four pixel points and the point P is large, the point P is an angular point; otherwise, the decision point P is not a corner point.
The core idea of the BRIEF algorithm is to select M point pairs in a certain mode around a pixel point P and combine comparison results of the M point pairs to serve as a descriptor. The specific implementation process is as follows:
taking the pixel point P as the center of a circle and d as the radius to make a circle; m point pairs are selected in the circle. In the embodiment, the value of M is 4, and an implementer selects the value of M according to the situation in practical application; the selected 4 point pairs were labeled as: p1(A,B)、P2(A,B)、P3(A,B)、P4(A, B), comparing the gray values of two points in each point pair, and describing the formula as follows:
wherein IARepresenting the gray value, I, of the first point, point A, of a pair of pointsBRepresenting the gray value of the second point in the pair, point B.
According to the description formula, the final descriptor form is: 1011.
the embodiment uses GMS algorithm to screen correct matching point pairs, GMS is a solution based on statistics, and can quickly distinguish correct matching and wrong matching, thereby improving the matching stability; the core idea is as follows: the number of pairs of correct matching points in the vicinity of correctly matching feature points should be greater than the number of pairs of correct matching points in the vicinity of incorrectly matching feature points, in terms of motion smoothness.
The GMS algorithm mainly comprises the following processes: detecting characteristic points and a calculation descriptor of an image subjected to image splicing; matching through a violence matching algorithm; dividing the image into G grids; and calculating the correct matching number near the feature points subjected to violent matching, and judging whether the points are correctly matched or not according to the correct matching number.
And solving a projection matrix between the images according to the obtained correct matching point pairs, projecting all input images onto a composite panoramic plane, calculating the coordinate range of the deformed images of the input images to obtain the size of an output image, calculating the offset between the original point of the source image and the original point of the output panoramic image, and then mapping the pixel of each input source image onto the output plane. And finally, fusing the images to eliminate gaps and the like in the spliced images, wherein a feather method is adopted in the embodiment, overlapped pixels are fused by using weighted average color values, and a pyramid fusion algorithm, a gradient fusion algorithm and the like can also be adopted by an implementer.
CIM of roads is based on city information data, and establishes a three-dimensional city space model and an organic complex of city information, including distribution information of isolation guardrails, abnormal monitoring information of the isolation guardrails, geographical position information and the like.
Performing semantic segmentation operation on the acquired image based on a semantic perception network, and perceiving and segmenting the isolation guardrail to obtain a guardrail semantic segmentation image; specifically, the method comprises the following steps:
the semantic perception network comprises a semantic segmentation encoder and a semantic segmentation decoder, and images containing isolation guardrails in roads are selected as training data sets, wherein 80% of the data sets are randomly selected as training sets, and the rest 20% of the data sets are selected as verification sets; labeling the data set, wherein the labeled road is 1, the isolation guardrail is 2, and the other classes are 3; the semantic aware network is trained using a cross entropy loss function.
The method comprises the steps of carrying out distortion removal processing on an acquired road isolation guardrail image, sending the road isolation guardrail image into a semantic perception network, obtaining a feature map through a semantic segmentation encoder, and processing the feature map through a semantic segmentation decoder to obtain a guardrail semantic segmentation map.
Operating the semantic guardrail segmentation graph based on the pixel classification network, acquiring pixel points which are judged as isolation guardrails in each row of pixels to obtain a guardrail scatter diagram, and fitting the scatter diagram by adopting a RANSAC method to obtain an isolation guardrail curve; wherein:
selecting semantic segmentation graphs comprising roads, isolation guardrails and other classes as training data sets, wherein 80% of the data sets are randomly selected as the training sets, and the rest 20% of the data sets are selected as verification sets; marking each row of isolation guardrails as a curve with the width of 1 pixel, specifically, marking the central point of the pixel corresponding to the isolation guardrail in each row of pixels as 1, and marking the other points as 0; training a pixel classification network by using a cross entropy loss function;
the pixel classification network comprises a classification encoder and a full-connection layer, wherein the input of the classification encoder is a guardrail semantic segmentation graph, the classification encoder is used for inputting the guardrail semantic segmentation graph and classifying pixels according to lines into the full-connection layer after extracting features, each line of pixels in the image respectively corresponds to one full-connection layer, and one full-connection layer is used for judging whether pixel points in one line of pixels are isolation guardrails or not.
The method for fitting the scatter diagram by adopting the RANSAC method comprises the following specific steps: randomly selecting N points in the scatter diagram to fit into a curve, calculating the distance from all the points to the curve, setting a threshold value, judging the points with the distance less than the threshold value as belonging to the curve, and recording the number of the points belonging to the curve; then randomly selecting N points, fitting a new curve again, and calculating the number of the points belonging to the curve according to the steps; finally, one curve in which the number of points belonging to the curve is the largest is selected as the curve of the barrier fence.
Image splicing is carried out on the images including the curves of the isolation guardrails according to the obtained projection change matrix, then the images are projected into the CIM to be combined with the panoramic image of the road in the CIM, and the CIM is subjected to visualization processing by utilizing a Web GIS technology, so that the real-time monitoring of the isolation guardrails is realized; the method comprises the steps of judging the movement and deformation degree of a guardrail according to the outlier condition of each isolation guardrail pixel scatter point, specifically, setting a distance threshold, judging that the isolation guardrail at the position is abnormal when the distance between the offset scatter point outside the isolation guardrail curve and the curve is larger than the threshold, and displaying warning information on a CIM visual interface when the isolation guardrail is detected to be abnormal so that a worker can maintain the abnormal isolation guardrail.
The above description is intended to provide those skilled in the art with a better understanding of the present invention and is not intended to limit the present invention.
Claims (4)
1. A road guardrail abnormity detection method based on artificial intelligence and image processing is characterized by comprising the following steps:
the method comprises the following steps that firstly, the acquired road isolation guardrail images are subjected to distortion removal processing and then are spliced, and then the images are projected into a CIM of a road which is constructed in advance;
performing semantic segmentation operation on the acquired image based on a semantic perception network, and perceiving and segmenting the isolation guardrail to obtain a guardrail semantic segmentation image;
selecting semantic segmentation graphs containing roads, isolation guardrails and other classes as a training data set; marking the isolation guardrails in the training data set as a curve with the width of 1 pixel; training a pixel classification network by using a cross entropy loss function;
operating the guardrail semantic segmentation graph based on the pixel classification network, acquiring pixel points which are judged as isolation guardrails in each row of pixels to obtain a guardrail scatter diagram, and fitting the scatter diagram to obtain an isolation guardrail curve; the pixel classification network comprises a classification encoder and a full-connection layer, wherein the input of the classification encoder is a guardrail semantic segmentation graph, and the guardrail semantic segmentation graph is sent into the full-connection layer to classify pixels after characteristics are extracted, wherein each row of pixels in the image respectively corresponds to one full-connection layer, and one full-connection layer is used for judging whether pixel points in one row of pixels are isolation guardrails or not;
fourthly, the images including the curves of the isolation guardrails are subjected to image splicing and then projected into the CIM, the CIM is subjected to visualization processing by utilizing a Web GIS technology, and when the isolation guardrails are detected to be abnormal, warning information is displayed on a CIM visualization interface; wherein, the unusual detection's of isolation barrier concrete mode does: and setting a distance threshold value, and judging that the isolation guardrail is abnormal when the distance between the deviation scatter outside the curve of the isolation guardrail and the curve is greater than the threshold value.
2. The method of claim 1, wherein the semantic aware network includes a semantic segmentation encoder and a semantic segmentation decoder, selecting an image in the road containing a barrier fence as a training data set; labeling the data set, wherein the labeled road is 1, the isolation guardrail is 2, and the other classes are 3; training the network using a cross entropy loss function; the method comprises the steps that a feature map is obtained through a semantic segmentation encoder after a collected road isolation guardrail image is subjected to distortion removal processing, and a semantic segmentation decoder processes the feature map to obtain a guardrail semantic segmentation map.
3. The method of claim 1, wherein fitting the scatter plot comprises: randomly selecting N points in the scatter diagram to fit into a curve, calculating the distance from all the points to the curve, setting a threshold value, judging the points with the distance less than the threshold value as belonging to the curve, and recording the number of the points belonging to the curve; then randomly selecting N points, fitting a new curve again, and calculating the number of the points belonging to the curve according to the steps; finally, the curve with the largest number of points belonging to the curve is selected as the curve of the isolation barrier.
4. The method of claim 1, wherein the step of image stitching comprises: and after extracting the characteristic points, matching the characteristic points, screening correct matching point pairs to obtain a projection matrix, and performing image deformation and image fusion operation after performing registration operation according to the projection matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010677433.6A CN111797803A (en) | 2020-07-15 | 2020-07-15 | Road guardrail abnormity detection method based on artificial intelligence and image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010677433.6A CN111797803A (en) | 2020-07-15 | 2020-07-15 | Road guardrail abnormity detection method based on artificial intelligence and image processing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111797803A true CN111797803A (en) | 2020-10-20 |
Family
ID=72807042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010677433.6A Withdrawn CN111797803A (en) | 2020-07-15 | 2020-07-15 | Road guardrail abnormity detection method based on artificial intelligence and image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111797803A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112257723A (en) * | 2020-10-26 | 2021-01-22 | 武汉中海庭数据技术有限公司 | Confidence evaluation method and system for guardrail extraction |
| CN112883948A (en) * | 2021-05-06 | 2021-06-01 | 深圳市城市交通规划设计研究中心股份有限公司 | Semantic segmentation and edge detection model building and guardrail abnormity monitoring method |
| CN114973138A (en) * | 2022-06-02 | 2022-08-30 | 松立控股集团股份有限公司 | Road surface abnormal object detection method based on high-order camera |
| CN116659540A (en) * | 2023-08-01 | 2023-08-29 | 西安博康硕达网络科技有限公司 | Traffic guardrail identification method in automatic driving process |
| CN116681955A (en) * | 2023-07-31 | 2023-09-01 | 深圳鲲云信息科技有限公司 | Method and computing device for identifying traffic guardrail anomalies |
-
2020
- 2020-07-15 CN CN202010677433.6A patent/CN111797803A/en not_active Withdrawn
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112257723A (en) * | 2020-10-26 | 2021-01-22 | 武汉中海庭数据技术有限公司 | Confidence evaluation method and system for guardrail extraction |
| CN112257723B (en) * | 2020-10-26 | 2022-09-27 | 武汉中海庭数据技术有限公司 | Confidence evaluation method and system for guardrail extraction |
| CN112883948A (en) * | 2021-05-06 | 2021-06-01 | 深圳市城市交通规划设计研究中心股份有限公司 | Semantic segmentation and edge detection model building and guardrail abnormity monitoring method |
| CN112883948B (en) * | 2021-05-06 | 2021-09-03 | 深圳市城市交通规划设计研究中心股份有限公司 | Semantic segmentation and edge detection model building and guardrail abnormity monitoring method |
| CN114973138A (en) * | 2022-06-02 | 2022-08-30 | 松立控股集团股份有限公司 | Road surface abnormal object detection method based on high-order camera |
| CN114973138B (en) * | 2022-06-02 | 2024-03-29 | 松立控股集团股份有限公司 | Road surface abnormal object detection method based on high-order camera |
| CN116681955A (en) * | 2023-07-31 | 2023-09-01 | 深圳鲲云信息科技有限公司 | Method and computing device for identifying traffic guardrail anomalies |
| CN116681955B (en) * | 2023-07-31 | 2023-11-28 | 深圳鲲云信息科技有限公司 | Method and computing device for identifying traffic guardrail anomalies |
| CN116659540A (en) * | 2023-08-01 | 2023-08-29 | 西安博康硕达网络科技有限公司 | Traffic guardrail identification method in automatic driving process |
| CN116659540B (en) * | 2023-08-01 | 2023-10-27 | 西安博康硕达网络科技有限公司 | Traffic guardrail identification method in automatic driving process |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111797803A (en) | Road guardrail abnormity detection method based on artificial intelligence and image processing | |
| AU2021285832B2 (en) | Method for property feature segmentation | |
| JP6180482B2 (en) | Methods, systems, products, and computer programs for multi-queue object detection and analysis (multi-queue object detection and analysis) | |
| WO2020000251A1 (en) | Method for identifying video involving violation at intersection based on coordinated relay of video cameras | |
| KR102197946B1 (en) | object recognition and counting method using deep learning artificial intelligence technology | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN110619279B (en) | Road traffic sign instance segmentation method based on tracking | |
| CN111967393A (en) | Helmet wearing detection method based on improved YOLOv4 | |
| CN106778688B (en) | Method for detecting crowd flow abnormal event in monitoring video in crowded scene | |
| CN102521842B (en) | Method and device for detecting fast movement | |
| CN113112480B (en) | Video scene change detection method, storage medium and electronic device | |
| CN106778633B (en) | Pedestrian identification method based on region segmentation | |
| CN112883948B (en) | Semantic segmentation and edge detection model building and guardrail abnormity monitoring method | |
| CN116824859B (en) | Intelligent traffic big data analysis system based on Internet of things | |
| CN112084928A (en) | Road traffic accident detection method based on visual attention mechanism and ConvLSTM network | |
| CN112084892A (en) | Road abnormal event detection management device and method thereof | |
| Bu et al. | A UAV photography–based detection method for defective road marking | |
| CN113947714B (en) | Multi-mode collaborative optimization method and system for video monitoring and remote sensing | |
| US11769238B2 (en) | Systems and methods for creating and/or analyzing three-dimensional models of infrastructure assets | |
| Hongquan et al. | Video scene invariant crowd density estimation using geographic information systems | |
| CN111797799A (en) | Subway passenger waiting area planning method based on artificial intelligence | |
| CN103903269B (en) | The description method and system of ball machine monitor video | |
| Fangfang et al. | Real-time lane detection for intelligent vehicles based on monocular vision | |
| CN105740819A (en) | Integer programming based crowd density estimation method | |
| CN112329671B (en) | Pedestrian running behavior detection method based on deep learning and related components |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20201020 |