CN104091344A - Road dividing method - Google Patents
Road dividing method Download PDFInfo
- Publication number
- CN104091344A CN104091344A CN201410350481.9A CN201410350481A CN104091344A CN 104091344 A CN104091344 A CN 104091344A CN 201410350481 A CN201410350481 A CN 201410350481A CN 104091344 A CN104091344 A CN 104091344A
- Authority
- CN
- China
- Prior art keywords
- estimator
- road
- link weights
- weights
- link
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000013507 mapping Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 30
- 230000008569 process Effects 0.000 claims description 9
- 238000005520 cutting process Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 7
- 230000006870 function Effects 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 230000008447 perception Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000004642 transportation engineering Methods 0.000 description 1
Landscapes
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a road dividing method. The method comprises the steps that S1, an n-link weight estimator is trained offline, the n-link weight estimator is obtained, a t-link weight estimator is trained offline, and the t-link weight estimator is obtained; S2, the n-link weight and the t-link weight are estimated through the n-link weight estimator and the t-link weight estimator respectively, a road image is divided online under an image division method framework, and a road area in the road image is obtained. According to the road dividing method, the mapping relation between the characteristics of adjacent pixel points and the n-link weight and the t-link weight is fitted with the statistical learning method, the trained n-link weight estimator and the trained t-link weight estimator are utilized for estimating the n-link weight and the t-link weight, the minimum division of a s-t image is obtained with the maximum flow/minimum division algorithm, and the road area is obtained. The road dividing method has good robustness for interference factors in the road environment.
Description
Technical field
The present invention relates to vehicle assistant drive technical field, particularly a kind of road dividing method for road detection system.
Background technology
The development of auto industry has had the history of more than 100 year, and the appearance of automobile has improved people's production and life style greatly, has promoted socioeconomic development.Automobile has also brought problems in offering convenience for human lives, and such as traffic hazard takes place frequently, energy resource consumption is too fast, and environmental pollution is serious etc.In the urban district of traffic congestion, the average per minute of driver need to complete 20-30 hand-foot coordination moves to realize the manipulation of vehicle, and therefore under traffic congestion situation, driving task is quite complicated.Along with social development, people to automobile the requirement in security, the feature of environmental protection, economy and comfortableness also more and more higher.The technology such as sensor, computing machine, control automatically, artificial intelligence and machine vision are in constantly development and innovation, and more and more application in Communication and Transportation Engineering.This background has expedited the emergence of the research and development of intelligent vehicle, makes vehicle possess autonomous driving and auxiliary function of driving, and experiences to realize safe, energy-conservation, convenient, comfortable driving.
Road detection system is the important component part of intelligent vehicle DAS (Driver Assistant System).Intelligent vehicle obtains position and the attitude of can pass through region and the relative road boundary of vehicle body by road detection system.So, can realize assisting vehicle navigation, lane departure warning, lane keeping, adaptive cruise, monitoring driver's driving condition, many auxiliary driving functions such as prediction driver's behavior intention by Road Detection.
Current most Approach for road detection is based on computer vision.In these methods, the process that road area pixel and other area pixel point are separated is referred to as road and cuts apart.It is a challenging problem that road is cut apart: on the one hand, due to the impact of the factors such as road surface material, weather condition, illumination variation, road surface has various outward appearances; On the other hand, along with the motion of vehicle, road surface and background be in dynamic change, and on road surface, generally exist the disturbing factor such as vehicle and pedestrian.The accuracy that above-mentioned factor is easily cut apart road impacts, and cuts apart and has brought very big difficulty for road.
Summary of the invention
(1) want technical solution problem
The object of the invention is to solve current road dividing method and be easily subject to the technical matters that environmental factor is disturbed, therefore, the present invention proposes a kind of road dividing method of robust.
(2) technical scheme
For solving the problems of the technologies described above, the present invention proposes a kind of road dividing method, comprises that step is as follows:
Step S1: off-line training n-link weights estimator, obtain n-link weights estimator, off-line training t-link weights estimator, obtains t-link weights estimator;
Step S2: utilize n-link weights estimator and t-link weights estimator to estimate respectively n-link weights and t-link weights, under the method frame cutting at figure, online road image is cut apart, obtain the road area in road image.
(3) beneficial effect
Under the method frame that the present invention cuts at figure, carry out road and cut apart, closely combined global information and the local message in road image, local interference is had to good robustness; The present invention carrys out the mapping relations between feature and the n-link weights of matching neighbor pixel by statistical learning method, the mode that can make up the contrast calculating n-link weights that use neighbor pixel in traditional method of cutting based on figure is subject to the shortcoming that environmental factor is disturbed, and has further strengthened the robustness that road is cut apart.Therefore, the present invention has good robustness to the disturbing factor in road environment.
Brief description of the drawings
Fig. 1 is the process flow diagram that road of the present invention is cut apart;
Fig. 2 is the connection diagram of adjacent node in s-t figure of the present invention;
Fig. 3 is s-t figure connection diagram of the present invention.
Embodiment
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in more detail.
The present invention proposes the figure that a kind of n-link weights of cutting apart for road can learn and cut method, road image is represented with s-t figure (s-t Graph), cut at figure under the method frame of (Graph Cut) and realize Road image segmentation, as shown in Figure 1, described method comprises:
Step S1: off-line training n-link (Neighboring Link) weights estimator, obtain n-link weights estimator, off-line training t-link (Terminal Link) weights estimator, obtains t-link weights estimator; N-link weights estimator is to carry out the feature of matching neighbor pixel to the mapping relations of n-link weights with statistical learning method.T-link weights estimator is the mapping relations that arrive t-link weights by the neighborhood characteristics of statistical learning method matching pixel.
Step S2: utilize n-link weights estimator and t-link weights estimator, estimate respectively n-link weights and t-link weights, under the method frame cutting at figure, online road image is cut apart, obtain the road area in road image.
It is a kind of combined optimization method based on graph theory that described figure cuts, and has a wide range of applications at computer vision field.When figure cuts when being applied to image and cutting apart, characterize image to be split with s-t figure, use max-flow/minimal cut (Max-flow/Min-cut) algorithm to ask for the minimal cut of s-t figure.The process of asking for minimal cut is equal to the process that minimizes an energy function, obtains minimal cut and can obtain the optimum segmentation of image under this energy function.
Described n-link weights estimator training process comprises the following steps:
Step S1A: extract training sample from the road image sample set marking;
Because pixel in road image is numerous, there is no need to extract all pixels in the image that marked as training sample.The sharp increase of on the one hand, extracting too much sample and can bring calculated amount; On the other hand, too much sample can make n-link weights estimator be difficult to focus on some disturbing factors, such as shade, and hot spot, lane markings etc.So, need to adopt a kind of sampling policy to extract sample from the image marking.A positive sample pixel adjacent from region, road surface, negative sample extracts the boundary from road.Not extracting positive sample from non-road area is because we are indifferent to the n-link weights of non-road area, and the diversity of minimizing data is conducive to the convergence of n-link weights estimator.In the present invention, according to certain probability, high-contrast area, high-brightness region and the general area from road area randomly drawed positive sample, the probability of institute's foundation has determined the ratio of each area sample in positive sample, and then has determined the power of n-link weights estimator for various different road conditions adaptive facultys.Pixel on road boundary is all used as negative sample, because the number of these pixels is far smaller than the number of road area pixel.The estimator output weights that positive sample is corresponding are 1, and the estimator output weights that negative sample is corresponding are 0.Each training sample extracting is the proper vector of one 10 × 1 dimension, included feature and define as shown in table 1.
The input feature vector of table 1 multi-layer perception（MLP）
Step S1B: utilize the training sample training n-link weights estimator extracting, until the error of n-link estimator, in receptible scope, obtains the parameter of n-link weights estimator.
Adopt multi-layer perception（MLP） (Multilayer Perceptron, MLP) to learn the mapping relations between feature and the n-link weights of neighbor pixel, for estimating n-link weights.The present invention represents road image with s-t figure, the adjacent node in s-t figure is connected by the mode of eight connections, and as shown in (a) in Fig. 2, adjacent node is by level, be vertically connected with the limit at diagonal angle.And in Fig. 2, (a) can be divided into multiple elementary cells that are made up of a node and four edges, as shown in (b) in Fig. 2.Certainly, there is no described four edges in the node of image border, but the limit lacking can be regarded as to weights are 0 limit.So, can with four independently MLP carry out the mapping relations of the weights of the n-link of the four direction shown in (b) in fitted figure 2.Each MLP is the feedforward neural network of four layers, comprises an input layer, and two hidden layers and an output layer comprise respectively 10,20,20,1 neurons.Each neuronic activation function is sigmoid function.The proper vector of one 10 × 1 dimension of input of each MLP, its definition is as shown in table 1.MLP output area is between 0 to 1, and output is more close to 1, and corresponding n-link weights are larger.
By RPROP algorithm, utilize the training sample training multi-layer perception（MLP） extracting in step S1A.For the impact on training result of the difference of eliminating positive sample and negative sample quantity, balance is hit rate and false alarm rate, need in the time of training, be that negative sample adds weight, and weight coefficient is as follows:
Wherein, w (y) is the weight coefficient adding for negative sample, N
pand N
nrepresent respectively the number of positive sample and negative sample, y is the class label of sample, and y=1 is positive sample, otherwise is negative sample.
The training process of described t-link weights estimator comprises the following steps:
Step S1a: extract training sample from the road image sample set marking;
Extract positive sample and negative sample according to certain ratio from road area and non-road area.Each sample is the proper vector of one 13 × 1 dimension, the feature comprising has R, G, the B value of pixel, average and the variance of R, G, B value in 9 × 9 windows centered by this pixel, the Grad of this pixel and gradient direction, and its coordinate under image coordinate system.Positive output valve corresponding to sample is 1, and the output valve that negative sample is corresponding is-1.
Step S1b: utilize the training sample training t-link weights estimator extracting, until the error of t-link weights estimator, in receptible scope, obtains the parameter of t-link weights estimator.
Adopt Gentle Adaboost (GAB) sorter as t-link weights estimator.Because GAB is a mutation algorithm of Adaboost algorithm, there is good feature selecting ability and numerical stability, have better robustness for noise, and GAB is a kind of good homing method that has supervision.This GAB sorter is made up of 500 decision-making stubs, and each decision-making stub can be regarded as a decision tree that only has a node.The output of GAB is not class label, but the weighted sum of the votes of all Weak Classifiers.For positive sample, we wish GAB output on the occasion of; For negative sample, we wish GAB output negative value.Utilize the training sample training GAB sorter extracting in step S1a.
The online cutting procedure of road image comprises the following steps:
Step S21: obtain road image I by vehicle-mounted imageing sensor;
Step S22: build s-t figure, each pixel in road image I is made as to adjacent node (Neighborhood Nodes), adjacent node is communicated with by four or eight modes that are communicated with are connected with adjacent node, source node (Source Node) represents road area and is made as prospect, sink nodes (Sink Node) represent non-road area and are made as background, and each adjacent node is connected with sink nodes with source node respectively; S-t figure is a kind of graph model, comprises two kinds of nodes and two kinds of limits.Two kinds of nodes are respectively adjacent node and end node, and end node comprises source node and convergence point.
As shown in Figure 3, what s indicated is source node to connection in s-t figure between node, and what t indicated is sink nodes, and solid black point is adjacent node.Eight mode of communicating that adjacent node passes through are connected, weights on every limit represent the correlation degree of these limit two endvertexs, the weights on the limit that adjacent node is connected with source node represent that adjacent node is the probability of the pixel of road area, and the weights on the limit that adjacent node is connected with sink nodes represent that adjacent node is the probability of non-road area pixel.
Step S23: utilize the n-link weights estimator that trains to estimate that respectively each adjacent node in road image I is adjacent the weights of the fillet between node;
Each adjacent node need to be estimated the You Qi left side, the limit adjacent node of weights, upper left side adjacent node, and top adjacent node and upper right side adjacent node fillet, as shown in (b) in Fig. 2.There is no described four edges in the node of image border, but the limit lacking can be regarded as to weights are 0 limit.The feature (defining as shown in table 1) of extracting each adjacent node, is input to feature in corresponding MLP.The output of MLP is the weights of corresponding sides, that is:
B
(p，q)＝MLP
d(v
p，q)，d＝dir(p，q) (2)
Wherein, B
(p, q)multi-layer perception（MLP） MLP
d(v
p, q) weights of corresponding sides of output, MLP
d(v
p, q) represent for estimating the multi-layer perception（MLP） of d the n-link weights in direction, v
p, qbe the proper vector of input, dir (p, q) represents the direction of neighbor p and q fillet.
Step S24: utilize the t-link weights estimator that trains to estimate respectively the weights of each adjacent node and the weights of source node fillet, each adjacent node and sink nodes fillet in road image I;
Extract the feature of each adjacent node, the GAB sorter that input trains.GAB sorter is exported the weighted sum of the votes of all Weak Classifiers, also needs further to pass through change of scale to interval [0,1], and the weight calculation of fillet is as follows:
Wherein, l
pfor the class label of pixel p, w is normal number, x
pthe output of GAB at pixel p place, l
p=1 o'clock R
p(x
p, l
p) for pixel p is to the weight of the fillet of sink nodes, l
p=0 o'clock R
p(x
p, l
p) for pixel p is to the weight of the fillet of source node.
Step S25: utilize max-flow/minimal cut algorithm to ask for the minimal cut of the s-t figure of structure, obtain the class label L={l of each pixel in road image I
i| l
i∈ 0,1}, and i=1,2 ... N}, wherein, N is the number of pixel in road image I, l
ifor the class label of pixel i, l
i=1 represent pixel point i is road area pixel, and no person is non-road area pixel.
Above-described specific embodiment; object of the present invention, technical scheme and beneficial effect are further described; institute is understood that; the foregoing is only specific embodiments of the invention; be not limited to the present invention; within the spirit and principles in the present invention all, any amendment of making, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.
Claims (7)
1. a road dividing method, comprises that step is as follows:
Step S1: off-line training n-link weights estimator, obtain n-link weights estimator, off-line training t-link weights estimator, obtains t-link weights estimator;
Step S2: utilize n-link weights estimator and t-link weights estimator to estimate respectively n-link weights and t-link weights, under the method frame cutting at figure, online road image is cut apart, obtain the road area in road image.
2. road dividing method as claimed in claim 1, is characterized in that, described road image represents with s-t figure.
3. road dividing method as claimed in claim 1, is characterized in that, described n-link weights estimator is to carry out the feature of matching neighbor pixel to the mapping relations of n-link weights with statistical learning method.
4. road dividing method as claimed in claim 1, is characterized in that, the process of described training n-link weights estimator comprises the following steps:
Step S1A: extract training sample from the road image sample set marking;
Step S1B: utilize the training sample training n-link weights estimator extracting, until the error of n-link estimator, in receptible scope, obtains the parameter of n-link weights estimator.
5. road dividing method as claimed in claim 1, is characterized in that, described t-link weights estimator is the mapping relations that arrive t-link weights by the neighborhood characteristics of statistical learning method matching pixel.
6. road dividing method as claimed in claim 1, is characterized in that, the process of described training t-link weights estimator comprises the following steps:
Step S1a: extract training sample from the road image sample set marking;
Step S1b: utilize the training sample training t-link weights estimator extracting, until the error of t-link weights estimator, in receptible scope, obtains the parameter of t-link weights estimator.
7. road dividing method as claimed in claim 1, is characterized in that, the described process of online road image being cut apart comprises the following steps:
Step S21: obtain road image by vehicle-mounted imageing sensor;
Step S22: build s-t figure, each pixel in road image is made as to adjacent node, adjacent node is communicated with by four or eight modes that are communicated with are connected with adjacent node, source node represents road area and is made as prospect, sink nodes represent non-road area and are made as background, and each adjacent node is connected with sink nodes with source node respectively;
Step S23: utilize the n-link weights estimator that trains to estimate that respectively each adjacent node in road image I is adjacent the weights of the fillet between node;
Step S24: utilize the t-link weights estimator that trains to estimate respectively the weights of each adjacent node and the weights of source node fillet, each adjacent node and sink nodes fillet in road image;
Step S25: utilize max-flow/minimal cut algorithm to ask for the minimal cut of the s-t figure of structure, obtain the class label L={l of each pixel in road image
i| l
i∈ 0,1}, and i=1,2 ... N}, wherein, N is the number of pixel in road image, l
ifor the class label of pixel i, l
i=1 represent pixel point i is road area pixel, otherwise is non-road area pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410350481.9A CN104091344B (en) | 2014-07-22 | 2014-07-22 | Road dividing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410350481.9A CN104091344B (en) | 2014-07-22 | 2014-07-22 | Road dividing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104091344A true CN104091344A (en) | 2014-10-08 |
| CN104091344B CN104091344B (en) | 2017-04-19 |
Family
ID=51639059
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410350481.9A Active CN104091344B (en) | 2014-07-22 | 2014-07-22 | Road dividing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104091344B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105631880A (en) * | 2015-12-31 | 2016-06-01 | 百度在线网络技术（北京）有限公司 | Lane line segmentation method and apparatus |
| CN106558058A (en) * | 2016-11-29 | 2017-04-05 | 北京图森未来科技有限公司 | Parted pattern training method, lane segmentation method, control method for vehicle and device |
| CN108229274A (en) * | 2017-02-28 | 2018-06-29 | 北京市商汤科技开发有限公司 | Multilayer neural network model training, the method and apparatus of roadway characteristic identification |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030069494A1 (en) * | 2001-10-04 | 2003-04-10 | Marie-Pierre Jolly | System and method for segmenting the left ventricle in a cardiac MR image |
| US20050134587A1 (en) * | 1999-09-23 | 2005-06-23 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
| US20050238215A1 (en) * | 2001-10-04 | 2005-10-27 | Marie-Pierre Jolly | System and method for segmenting the left ventricle in a cardiac image |
| CN101558404A (en) * | 2005-06-17 | 2009-10-14 | 微软公司 | Image segmentation |
| CN102609686A (en) * | 2012-01-19 | 2012-07-25 | 宁波大学 | Pedestrian detection method |
| CN103473767A (en) * | 2013-09-05 | 2013-12-25 | 中国科学院深圳先进技术研究院 | Segmentation method and system for abdomen soft tissue nuclear magnetism image |
-
2014
- 2014-07-22 CN CN201410350481.9A patent/CN104091344B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050134587A1 (en) * | 1999-09-23 | 2005-06-23 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
| US20030069494A1 (en) * | 2001-10-04 | 2003-04-10 | Marie-Pierre Jolly | System and method for segmenting the left ventricle in a cardiac MR image |
| US20050238215A1 (en) * | 2001-10-04 | 2005-10-27 | Marie-Pierre Jolly | System and method for segmenting the left ventricle in a cardiac image |
| CN101558404A (en) * | 2005-06-17 | 2009-10-14 | 微软公司 | Image segmentation |
| CN102609686A (en) * | 2012-01-19 | 2012-07-25 | 宁波大学 | Pedestrian detection method |
| CN103473767A (en) * | 2013-09-05 | 2013-12-25 | 中国科学院深圳先进技术研究院 | Segmentation method and system for abdomen soft tissue nuclear magnetism image |
Non-Patent Citations (4)
| Title |
|---|
| FALIU YI 等: "Image Segmentation: A Survey of Graph-cut Methods", 《SYSTEMS AND INFORMATICS》 * |
| JYUN-FAN TSAI 等: "Road Detection and Classification in Urban Environments Using Conditional Random Field Models", 《INTELLIGENT TRANSPORTATION SYSTEMS CONFERENCE》 * |
| SHA YUN 等: "A Road Detection Algorithm by Boosting Using Feature Combination", 《INTELLIGENT VEHICLES SYMPOSIUM》 * |
| YURI BOYKOV 等: "An Experimental Comparison of Min-Cut/Max-Flow Algorithms for Energy Minimization in Vision", 《PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105631880A (en) * | 2015-12-31 | 2016-06-01 | 百度在线网络技术（北京）有限公司 | Lane line segmentation method and apparatus |
| CN105631880B (en) * | 2015-12-31 | 2019-03-22 | 百度在线网络技术（北京）有限公司 | Lane line dividing method and device |
| CN106558058A (en) * | 2016-11-29 | 2017-04-05 | 北京图森未来科技有限公司 | Parted pattern training method, lane segmentation method, control method for vehicle and device |
| CN108229274A (en) * | 2017-02-28 | 2018-06-29 | 北京市商汤科技开发有限公司 | Multilayer neural network model training, the method and apparatus of roadway characteristic identification |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104091344B (en) | 2017-04-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Saleem et al. | Smart cities: Fusion-based intelligent traffic congestion control system for vehicular networks using machine learning techniques | |
| US11934962B2 (en) | Object association for autonomous vehicles | |
| Liang et al. | Fine-grained vessel traffic flow prediction with a spatio-temporal multigraph convolutional network | |
| CN107886073A (en) | A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks | |
| Yi et al. | Trajectory clustering aided personalized driver intention prediction for intelligent vehicles | |
| CN105893951A (en) | Multidimensional non-wearable type traffic police gesture identification method and system for driverless vehicles | |
| CN104537891B (en) | A kind of boats and ships track real-time predicting method | |
| Chandra et al. | Cmetric: A driving behavior measure using centrality functions | |
| CN111045422A (en) | Control method for automatically driving and importing 'machine intelligence acquisition' model | |
| CN103839264A (en) | Detection method of lane line | |
| CN111881802B (en) | Traffic police gesture recognition method based on double-branch space-time graph convolutional network | |
| CN104331160A (en) | Lip state recognition-based intelligent wheelchair human-computer interaction system and method | |
| CN104091344A (en) | Road dividing method | |
| Chen et al. | Robust vehicle detection and viewpoint estimation with soft discriminative mixture model | |
| Arefnezhad et al. | Modeling of double lane change maneuver of vehicles | |
| US11550327B2 (en) | Composition method of automatic driving machine consciousness model | |
| CN104050680A (en) | Image segmentation method based on iteration self-organization and multi-agent inheritance clustering algorithm | |
| Baliyan et al. | Role of AI and IoT techniques in autonomous transport vehicles | |
| Hoang et al. | Optimizing YOLO Performance for Traffic Light Detection and End-to-End Steering Control for Autonomous Vehicles in Gazebo-ROS2 | |
| Souza et al. | Vision-based waypoint following using templates and artificial neural networks | |
| CN105809699A (en) | Image segmentation based car window extraction method and system | |
| CN106650814A (en) | Vehicle-mounted monocular vision-based outdoor road adaptive classifier generation method | |
| Zailan et al. | An automated solid waste detection using the optimized YOLO model for riverine management | |
| CN111046897A (en) | Method for defining fuzzy event probability measure spanning different spaces | |
| CN114332833A (en) | Binary differentiable fatigue detection method based on face key points |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |