CN110517349A - A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint - Google Patents
A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint Download PDFInfo
- Publication number
- CN110517349A CN110517349A CN201910684070.6A CN201910684070A CN110517349A CN 110517349 A CN110517349 A CN 110517349A CN 201910684070 A CN201910684070 A CN 201910684070A CN 110517349 A CN110517349 A CN 110517349A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- target detection
- camera
- bounding box
- target
- 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
Classifications
-
- 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
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/97—Determining parameters from multiple pictures
-
- 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
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The 3D vehicle target detection method based on monocular vision and geometrical constraint that the invention discloses a kind of: by carrying out target detection to each frame image of camera video, vehicle 2D bounding box and type of vehicle 2D target detection step: are obtained；Vehicle observation angle obtaining step: vehicle observation angle is obtained by convolutional neural networks；Transverse and longitudinal coordinate obtaining step: length, width and height, camera inside and outside parameter and the target vehicle attitude parameter of the vehicle observation angle, target vehicle that are obtained by S2 obtain the transverse and longitudinal coordinate of 3D target vehicle bounding box subpoint in the picture.3D frame plot step: by the way that you draw 3D bounding box to the vehicle length, width and height and vehicle-posture information detected.For the present invention by combining 2D target detection, the information such as viewing angle and type of vehicle propose a kind of 3D vehicle target detection method, and finally provide the 3D bounding box and posture information of vehicle.
Description
Technical field
The present invention relates to pilotless target detection technique fields, more particularly to one kind to be based on monocular vision and geometrical constraint
3D vehicle target detection method.
Background technique
Although the 2D vehicle target detection algorithm of current many view-based access control models detects fast speed, it can not be more accurate
The posture and dimension information of ground offer vehicle.2D target detection, which is only used only, can lose many important geometrical characteristics；And big portion
The depth information for dividing 3D vehicle target detection method to be all based on LiDAR offer again.The 3D target detection of view-based access control model can mention
For the 3D target detection scheme more more inexpensive than LiDAR, however the 3D vehicle target detection method of many view-based access control models at present
Complicated neural network and prior model are needed, be difficult in small-sized Platform deployment or realizes faster detection speed.Therefore, originally
The 3D vehicle target detection method based on monocular vision that patent proposes, can be on the platform of small-sized or low computing capability
It realizes vehicle 3D target detection, that is, provides the information such as 2D, 3D bounding box and the type of vehicle of vehicle.
Summary of the invention
It is an object of the present invention in view of the above-mentioned problems, propose a kind of 3D vehicle based on monocular vision and geometrical constraint
Object detection method.
A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint, includes the following steps:
Image obtains: the frame image of target vehicle is obtained by vehicle-mounted monocular camera；
2D target detection: designing and training 2D target detection network, obtains the type of vehicle and 2D of target vehicle in image
Four apex coordinate (x of bounding boxmin,ymin,xmax,ymax)；
Viewing angle obtains: designing and training viewing angle estimates network, obtain the viewing angle of target vehicle；
3D bounding box is drawn: according to type of vehicle, viewing angle, 2D bounding box information and camera inside and outside parameter, being obtained
The transverse and longitudinal coordinate of 3D target vehicle bounding box subpoint in the picture draws 3D bounding box and simultaneously exports vehicle-posture information.
A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, institute
Viewing angle is stated by obtaining using convolutional neural networks, network training uses homemade viewing angle data set, data set
In every samples pictures size be 64*64 pixel, the viewing angle of sample label sample vehicle thus.
A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, institute
2D target detection is stated by one or more deep learning targets of YOLO, SSD and RetinaNet to detect, the weight of network
By focal loss, GIoU loss one or more kinds of loss functions in the case of by training obtain.
A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, institute
The camera coordinates system setting X-axis positive direction for stating vehicle-mounted camera is vehicle front-right, and Y-axis positive direction is immediately below vehicle, and Z axis is just
Direction is right ahead.
A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, institute
Stating 3D bounding box and drawing includes following sub-step:
S1: vehicle length, width and height and camera inside and outside parameter are obtained by tabling look-up；
S2: according to table look-up and the camera coordinates system of vehicle-mounted camera obtain vehicle bottom center in vehicle-mounted monocular camera Y
Axis direction translation distance value ty；
S3: the transverse and longitudinal coordinate of vehicle 3D bounding box subpoint in the picture is calculated；
S4: the inside and outside ginseng of the l for the vehicle tabled look-up, w, h and camera projects in the picture to represent vehicle 3D bounding box
The transverse and longitudinal coordinate of point, finds out (tx,tz,ry), wherein txExpression target vehicle is at a distance from the left and right directions of vehicle-mounted camera, tz
Indicate the distance in the front-back direction of target vehicle and vehicle-mounted camera, ryIndicate what target headstock was rotated around vehicle-mounted camera Y-axis
Angle；
S5: output vehicle-posture information (l, w, h, tx, ty,,tz,ry), carry out the drafting of 3D frame.
This reality advantageous effect of the invention: result (bounding box position, target type) lower of the present invention in 2D target detection
Increase an observation parameter and is achieved that 3D bounding box is estimated.It will be according to 6-9 freely not necessarily like other 3D algorithm of target detection
Parameter is spent just it can be concluded that bounding box；The complexity and calculation amount for not only reducing 3D target detection network, also reduce data
The complexity and calculation amount for reducing 3D target detection network for collecting production also reduce the difficulty of data set production.
Detailed description of the invention
Fig. 1 is the 3D bounding box of vehicle under different viewing angles and the relation schematic diagram of 2D bounding box；
Fig. 2 is the relation schematic diagram of the projection of viewing angle alpha, 2D bounding box and 3D bounding box；
Fig. 3 is the camera coordinates system definition figure of in-vehicle camera；
Fig. 4 is the implementation procedure flow chart of 3D vehicle target detection method of the invention；
Fig. 5 is the test effect using KITTI data set and practical campus road video interception
Specific embodiment
For a clearer understanding of the technical characteristics, objects and effects of the present invention, this hair of Detailed description of the invention is now compareed
Bright specific embodiment.
In the present embodiment, a kind of lane line real-time detection method based on FPGA includes the following steps:
S1:2D target detection step: the input of system is each frame image in vehicle-mounted monocular camera video stream, that is, is schemed
Vehicle as in is from front side vehicle.Input of each frame image as 2D target detection.2D target detection network exports
" type of vehicle (such as car, truck, middle bus, motorbus) describes four apex coordinate (x of 2D bounding boxmin,
ymin, xmax,ymax)”
S2: vehicle observation angle obtaining step: viewing angle estimates that network is convolutional neural networks, and input is 2D target
Image (pass through Resize to 64*64 pixel) in testing result in bounding box, export for viewing angle (it may range from-
180 °~180 °, or the forms such as 0-1, other angles system, Circular measure).Network training uses homemade viewing angle degree
According to collection, every samples pictures size is 64*64 pixel in data set, the viewing angle of sample label sample vehicle thus.
S3: posture information (l, w, h, the t of vehiclex,ty,tz,ry) obtain including following sub-step:
S31: the corresponding relationship of 3D and 2D bounding box are divided into four kinds of situations by angle according to the observation, obtain abscissa xmid, four
Kind situation is as shown in figure 1 shown in four subgraphs.The relationship for determining target carriage 3D bounding box and 2D bounding box is four kinds of situations in Fig. 1
One of after, length and width l, the w of viewing angle alpha and vehicle that projection relation, S2 step obtain according to Fig.2, can obtain Fig. 1
(a) the abscissa x of midpoint 3, the midpoint Fig. 1 (b) 4, the midpoint Fig. 1 (c) 1 or the midpoint Fig. 1 (d) 2 in the picturemid；
S32: it obtains and contains vehicle attitude unknown number (tx,tz,ry) the coordinate that projects in the picture of 3D bounding box angle point,
Wherein tri- axis direction of camera coordinates system X, Y, Z of in-vehicle camera and relationship such as Fig. 3 of automobile；
The type of vehicle obtained by 2D target detection searches the type of vehicle and vehicle of pre-production according to type of vehicle
It is flat in in-vehicle camera Y direction that size table obtains the approximate length, width and height l of this vehicle, w, h (unit is rice) and vehicle bottom center
Distance value ty is moved, the unknown number for concurrently setting vehicle attitude is (tx,tz,ry), that is, indicate (the left and right directions with vehicle-mounted camera
Distance, distance in the front-back direction, the angle that headstock is rotated around vehicle-mounted camera Y-axis), use this unknown number (tx,tz,ry), look into
The inside and outside ginseng of the l for the vehicle that table obtains, w, h and camera represents the transverse and longitudinal coordinate of vehicle 3D bounding box subpoint in the picture.
In case where Fig. 1 third figure, the coordinate of three key points is
4 abscissas of point:
1 abscissa of point:
1 ordinate of point:
Wherein, (x0,y0) be camera principle point location picture, (tx,ty,tz) it is the bottom center of 3D bounding box respectively in vehicle-mounted phase
Machine X, Y, Z axis three distances (unit is rice) just translated up, (fx,fy) it is the focal length of camera obtained by camera internal reference
Value；
S33: by taking the situation of Fig. 1 (c) as an example, joint type (1-1), (1-2), (1-3) and (xmax,ymax,xmid), obtain three
A equation, such as formula (1-4).Solving equations obtain (tx,tz,ry):
The processing method of other three kinds of situations in S34: Fig. 1, the case where figure is opened for Fig. 1 first, the group of simultaneous equations
Point is the cross of point 3, ordinate and the abscissa for putting 2, the case where opening figure for Fig. 1 second, and the point of Simultaneous Equations is a little 4
Horizontal, ordinate and put 1 abscissa, the case where for the 4th figure of Fig. 1, the point of Simultaneous Equations be a little 2 cross, ordinate and
The abscissa of point 3.
S4:3D frame plot step obtains posture information (l, w, h, the t of all vehicles to be detected in a frame imagex,ty,
tz,ry) after, and draw 2D/3D bounding box.The implementation procedure flow chart of entire 3D vehicle target detection method is as shown in Figure 4.Figure
5 first figure in the same direction sail vehicle camera left front, opposite direction sail come vehicle in camera left front, sail vehicle in the same direction
In camera right front；Second figure is that the vehicle that sails in the same direction sails that carry out vehicle left front in camera in camera left front, opposite direction
Side；Third figure is that the vehicle in the same direction that sails in camera left front, in the same direction sails vehicle in camera right front；4th figure be
The vehicle in the same direction that sails sails vehicle in camera right front in camera left front, in the same direction.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes
Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its
Equivalent thereof.
Claims (5)
1. a kind of 3D vehicle target detection method based on monocular vision and geometrical constraint, which comprises the steps of:
Image obtains: the frame image of target vehicle is obtained by vehicle-mounted monocular camera；
2D target detection: designing and training 2D target detection network, obtains the type of vehicle of target vehicle and the boundary 2D in image
Frame four apex coordinates (x min, y min, x max , y max)；
Viewing angle obtains: designing and training viewing angle estimates network, obtain the viewing angle of target vehicle；
3D bounding box is drawn: according to type of vehicle, viewing angle, 2D bounding box information and camera inside and outside parameter, obtaining 3D mesh
The transverse and longitudinal coordinate of vehicle bounding box subpoint in the picture is marked, 3D bounding box is drawn and simultaneously exports vehicle-posture information.
2. a kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, special
Sign is that for the viewing angle by obtaining using convolutional neural networks, network training uses homemade viewing angle data
Collection, every samples pictures size is 64*64 pixel in data set, the viewing angle of sample label sample vehicle thus.
3. a kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, special
Sign is that the 2D target detection is detected by one or more deep learning targets of YOLO, SSD and RetinaNet, net
The weight of network by focal loss, GIoU loss one or more kinds of loss functions in the case of by training obtain.
4. a kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, special
Sign is, the camera coordinates system setting X-axis positive direction of the vehicle-mounted camera is vehicle front-right, Y-axis positive direction be vehicle just
Lower section, Z axis positive direction are right ahead.
5. a kind of 3D vehicle target detection method based on monocular vision and geometrical constraint according to claim 1, special
Sign is that it includes following sub-step that the 3D bounding box, which is drawn:
S1: vehicle length, width and height and camera inside and outside parameter are obtained by tabling look-up；
S2: according to table look-up and the camera coordinates system of vehicle-mounted camera obtain vehicle bottom center in vehicle-mounted monocular camera Y-axis side
To translation distance value ty；
S3: the transverse and longitudinal coordinate of vehicle 3D bounding box subpoint in the picture is calculated；
S4: the inside and outside ginseng of the l for the vehicle tabled look-up, w, h and camera is to represent vehicle 3D bounding box subpoint in the picture
Transverse and longitudinal coordinate, find out (tx, tz, ry), wherein txExpression target vehicle is at a distance from the left and right directions of vehicle-mounted camera, tz
Indicate the distance in the front-back direction of target vehicle and vehicle-mounted camera, ry indicates what target headstock was rotated around vehicle-mounted camera Y-axis
Angle；
S5: output vehicle-posture information (l, w, h, tx, ty tz, ry), carry out the drafting of 3D frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910684070.6A CN110517349A (en) | 2019-07-26 | 2019-07-26 | A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910684070.6A CN110517349A (en) | 2019-07-26 | 2019-07-26 | A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110517349A true CN110517349A (en) | 2019-11-29 |
Family
ID=68623637
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910684070.6A Pending CN110517349A (en) | 2019-07-26 | 2019-07-26 | A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110517349A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369617A (en) * | 2019-12-31 | 2020-07-03 | 浙江大学 | 3D target detection method of monocular view based on convolutional neural network |
| CN111563525A (en) * | 2020-03-25 | 2020-08-21 | 北京航空航天大学 | Moving target detection method based on YOLOv3-Tiny |
| CN112016532A (en) * | 2020-10-22 | 2020-12-01 | 腾讯科技（深圳）有限公司 | Vehicle detection method and device |
| CN112183241A (en) * | 2020-09-11 | 2021-01-05 | 北京罗克维尔斯科技有限公司 | Target detection method and device based on monocular image |
| US20210209341A1 (en) * | 2020-05-15 | 2021-07-08 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Three-dimensional object detection method, electronic device and readable storage medium |
| CN113128434A (en) * | 2021-04-27 | 2021-07-16 | 南京大学 | Method for carrying out 3D target detection on monocular RGB image |
| CN113191329A (en) * | 2021-05-26 | 2021-07-30 | 超级视线科技有限公司 | Vehicle berth matching method and system based on monocular vision picture |
| WO2021217924A1 (en) * | 2020-04-29 | 2021-11-04 | 平安国际智慧城市科技股份有限公司 | Method and apparatus for identifying vehicle type at traffic checkpoint, and device and storage medium |
| CN114387202A (en) * | 2021-06-25 | 2022-04-22 | 南京交通职业技术学院 | 3D target detection method based on vehicle end point cloud and image fusion |
| WO2022246767A1 (en) * | 2021-05-27 | 2022-12-01 | 华为技术有限公司 | Method and device for determining steering intention of target vehicle |
| EP4105820A4 (en) * | 2020-03-06 | 2023-07-26 | Huawei Technologies Co., Ltd. | Method and device for acquiring 3d information of vehicle |
| CN116645649A (en) * | 2023-07-27 | 2023-08-25 | 深圳魔视智能科技有限公司 | Vehicle pose and size estimation method, device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107256633A (en) * | 2017-05-09 | 2017-10-17 | 西安理工大学 | A kind of vehicle type classification method based on monocular cam three-dimensional estimation |
| CN107407730A (en) * | 2015-03-06 | 2017-11-28 | 高通股份有限公司 | The real-time occupancy map creation system of autonomous vehicle |
| CN108759667A (en) * | 2018-05-29 | 2018-11-06 | 福州大学 | Front truck distance measuring method based on monocular vision and image segmentation under vehicle-mounted camera |
| CN108898628A (en) * | 2018-06-21 | 2018-11-27 | 北京纵目安驰智能科技有限公司 | Three-dimensional vehicle object's pose estimation method, system, terminal and storage medium based on monocular |
| CN109241856A (en) * | 2018-08-13 | 2019-01-18 | 浙江零跑科技有限公司 | A kind of vehicle-mounted vision system solid object detection method of monocular |
| CN109544633A (en) * | 2017-09-22 | 2019-03-29 | 华为技术有限公司 | Target ranging method, device and equipment |
-
2019
- 2019-07-26 CN CN201910684070.6A patent/CN110517349A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107407730A (en) * | 2015-03-06 | 2017-11-28 | 高通股份有限公司 | The real-time occupancy map creation system of autonomous vehicle |
| CN107256633A (en) * | 2017-05-09 | 2017-10-17 | 西安理工大学 | A kind of vehicle type classification method based on monocular cam three-dimensional estimation |
| CN109544633A (en) * | 2017-09-22 | 2019-03-29 | 华为技术有限公司 | Target ranging method, device and equipment |
| CN108759667A (en) * | 2018-05-29 | 2018-11-06 | 福州大学 | Front truck distance measuring method based on monocular vision and image segmentation under vehicle-mounted camera |
| CN108898628A (en) * | 2018-06-21 | 2018-11-27 | 北京纵目安驰智能科技有限公司 | Three-dimensional vehicle object's pose estimation method, system, terminal and storage medium based on monocular |
| CN109241856A (en) * | 2018-08-13 | 2019-01-18 | 浙江零跑科技有限公司 | A kind of vehicle-mounted vision system solid object detection method of monocular |
Non-Patent Citations (1)
| Title |
|---|
| ARSALAN MOUSAVIAN等: ""3D Bounding Box Estimation Using Deep Learning and Geometry"", 《PROCEEDINGS OF THE IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369617A (en) * | 2019-12-31 | 2020-07-03 | 浙江大学 | 3D target detection method of monocular view based on convolutional neural network |
| CN111369617B (en) * | 2019-12-31 | 2022-06-21 | 浙江大学 | 3D target detection method of monocular view based on convolutional neural network |
| EP4105820A4 (en) * | 2020-03-06 | 2023-07-26 | Huawei Technologies Co., Ltd. | Method and device for acquiring 3d information of vehicle |
| CN111563525A (en) * | 2020-03-25 | 2020-08-21 | 北京航空航天大学 | Moving target detection method based on YOLOv3-Tiny |
| WO2021217924A1 (en) * | 2020-04-29 | 2021-11-04 | 平安国际智慧城市科技股份有限公司 | Method and apparatus for identifying vehicle type at traffic checkpoint, and device and storage medium |
| US20210209341A1 (en) * | 2020-05-15 | 2021-07-08 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Three-dimensional object detection method, electronic device and readable storage medium |
| US11587338B2 (en) * | 2020-05-15 | 2023-02-21 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Three-dimensional object detection method, electronic device and readable storage medium |
| WO2022053015A1 (en) * | 2020-09-11 | 2022-03-17 | 北京罗克维尔斯科技有限公司 | Target detection method and apparatus based on monocular image |
| CN112183241A (en) * | 2020-09-11 | 2021-01-05 | 北京罗克维尔斯科技有限公司 | Target detection method and device based on monocular image |
| CN112016532B (en) * | 2020-10-22 | 2021-02-05 | 腾讯科技（深圳）有限公司 | Vehicle detection method and device |
| CN112016532A (en) * | 2020-10-22 | 2020-12-01 | 腾讯科技（深圳）有限公司 | Vehicle detection method and device |
| CN113128434A (en) * | 2021-04-27 | 2021-07-16 | 南京大学 | Method for carrying out 3D target detection on monocular RGB image |
| CN113128434B (en) * | 2021-04-27 | 2023-11-21 | 南京大学 | Method for carrying out 3D target detection on monocular RGB image |
| CN113191329A (en) * | 2021-05-26 | 2021-07-30 | 超级视线科技有限公司 | Vehicle berth matching method and system based on monocular vision picture |
| WO2022246767A1 (en) * | 2021-05-27 | 2022-12-01 | 华为技术有限公司 | Method and device for determining steering intention of target vehicle |
| CN114387202A (en) * | 2021-06-25 | 2022-04-22 | 南京交通职业技术学院 | 3D target detection method based on vehicle end point cloud and image fusion |
| CN116645649A (en) * | 2023-07-27 | 2023-08-25 | 深圳魔视智能科技有限公司 | Vehicle pose and size estimation method, device and storage medium |
| CN116645649B (en) * | 2023-07-27 | 2023-12-08 | 深圳魔视智能科技有限公司 | Vehicle pose and size estimation method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110517349A (en) | A kind of 3D vehicle target detection method based on monocular vision and geometrical constraint | |
| CN108759667B (en) | Front truck distance measuring method under vehicle-mounted camera based on monocular vision and image segmentation | |
| CN111429514B (en) | Laser radar 3D real-time target detection method integrating multi-frame time sequence point cloud | |
| CN108638999B (en) | Anti-collision early warning system and method based on 360-degree look-around input | |
| CN106462968B (en) | Method and device for calibrating a camera system of a motor vehicle | |
| CN106156723B (en) | A kind of crossing fine positioning method of view-based access control model | |
| CN110032949A (en) | A kind of target detection and localization method based on lightweight convolutional neural networks | |
| CN109636771B (en) | Flight target detection method and system based on image processing | |
| CN107633220A (en) | A kind of vehicle front target identification method based on convolutional neural networks | |
| CN113506318B (en) | Three-dimensional target perception method under vehicle-mounted edge scene | |
| CN108805906A (en) | A kind of moving obstacle detection and localization method based on depth map | |
| CN110969064B (en) | Image detection method and device based on monocular vision and storage equipment | |
| CN109685732A (en) | A kind of depth image high-precision restorative procedure captured based on boundary | |
| CN105225230A (en) | A kind of method and device identifying foreground target object | |
| CN107677274A (en) | Unmanned plane independent landing navigation information real-time resolving method based on binocular vision | |
| CN108009494A (en) | A kind of intersection wireless vehicle tracking based on unmanned plane | |
| CN112101160B (en) | Binocular semantic SLAM method for automatic driving scene | |
| CN115082881A (en) | Target detection method, storage medium, electronic device, and vehicle | |
| CN112489106A (en) | Video-based vehicle size measuring method and device, terminal and storage medium | |
| CN105976324A (en) | Vehicle image splicing method | |
| CN105300390A (en) | Method and device for determining moving trace of obstacle | |
| CN111462241A (en) | Target positioning method based on monocular vision | |
| CN110197104B (en) | Distance measurement method and device based on vehicle | |
| Yang et al. | Vision-based intelligent vehicle road recognition and obstacle detection method | |
| Zheng et al. | A robust strategy for roadside cooperative perception based on multi-sensor fusion |
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 |