Target detection method, device and storage medium
A target detection and target prediction technology, applied in the computer field, can solve the problems of disjoint regression function between predicted frame and real frame, slow network convergence speed, failure, etc.
Active Publication Date: 2021-04-27
知行汽车科技(苏州)股份有限公司
View PDF6 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] This application provides a target detection method, device and storage medium, which can solve the problem of slow network convergence and regression when the predicted frame and the real frame are disjoint or contained when only using the area coincidence degree information for position regression to train the target detection network. Problems with failure
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0057] The specific implementation manners of the present application will be further described in detail below in conjunction with the drawings and embodiments. The following examples are used to illustrate the present application, but not to limit the scope of the present application.
[0058] First, some terms involved in this application are introduced.
[0059] Self-driving: It is a smart car that realizes automatic driving through a computer system.
[0060] Point cloud data (Point Cloud): refers to a set of vectors in a three-dimensional coordinate system.
[0061] Neural Network: It is an algorithm model that imitates the behavior characteristics of animal neural networks and performs distributed parallel information processing.
[0062] Target Detection: It refers to finding out all the objects of interest in the image and determining their location and category.
[0063] Voxel Grid: It is a method that divides the three-dimensional space into a set of tiny three-d...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The present application relates to a target detection method, device and storage medium, belonging to the field of computer technology. The method includes: acquiring 3D data in the form of voxel grids corresponding to 3D point cloud data; inputting the 3D data into a 3D target detection network to obtain The target prediction results include the center point coordinates, length, width, height and orientation angle of the target prediction frame; obtain the corresponding labeling results of the three-dimensional data, and the labeling results include the center point coordinates, length, width, height and orientation angle of the target labeling frame; according to The labeling results and target prediction results determine the 3D rotation overlap rate, the offset between the center point coordinates, the diagonal distance between the target labeling frame and the bounding box of the target prediction frame; based on the target prediction results, labeling results, and 3D rotation overlap rate , offset and diagonal distance generate a loss function to train the three-dimensional target detection network to obtain the final target detection network; it can improve the network convergence speed and training efficiency.
Description
technical field [0001] The present application relates to a target detection method, device and storage medium, belonging to the technical field of computers. Background technique [0002] With the development of autonomous driving technology, the vehicle can collect 3D point cloud data through the lidar installed on it, and perform 3D target detection based on the 3D point cloud data. [0003] When performing 3D target detection based on 3D point cloud data, by inputting the 3D point cloud data into the preset neural network, and using the regression loss function of 3-dimension Intersection-over-Union (3D-IOU) to The neural network is trained to obtain the final target detection network, and the target detection network is used to detect the target in the three-dimensional point cloud data. [0004] However, 3D-IOU can only reflect the overlap rate between the target real value and the predicted value, so as to iteratively increase the overlap rate to reduce the loss. Ho...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G01S7/48G01S17/894
CPCG01S7/4802G01S17/894
Inventor 陈伟
Owner 知行汽车科技(苏州)股份有限公司