A Pedestrian Detection Method Based on Neural Network Multi-scale Feature Map
A multi-scale feature and pedestrian detection technology, applied in the field of image recognition, can solve the problem of low robustness
Active Publication Date: 2021-09-10
CHINACCS INFORMATION IND
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
At present, the traditional pedestrian detection method trains the classifier by extracting features such as HOG, LBP, and Haar. The obtained model has a good detection rate for pedestrians with small posture changes, but the robustness is not high.
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
[0033] In recent years, methods based on convolutional neural networks have made great achievements in the field of computer vision, such as: object detection, feature matching, pose estimation and many other tasks. In order to greatly improve the efficiency of existing pedestrian detection, the present invention proposes a pedestrian detection method based on neural network multi-scale feature maps, using the residual network as a feature extraction network, and splicing with the multi-scale feature map pedestrian detection network, which can Detection is performed in the case of complex scenes and large changes in pedestrian scales, and has the advantages of high accuracy and high detection efficiency, especially the detection speed can be increased by 4 times.
[0034] In order to clearly illustrate the technical features of the solution, the solution will be described below through specific implementation modes.
[0035] The embodiment of the present invention provides a p...
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 invention discloses a pedestrian detection method based on a multi-scale feature map of a neural network. The detection method is as follows: collecting pedestrian detection samples; performing frame truncating processing on selected videos, and only retaining pictures containing pedestrians; making sample labels; building lightweight Residual feature extraction network; building a multi-scale feature map pedestrian detection network; performing network pre-training on the built feature extraction network in the sample data set; splicing the trained feature extraction network with the built detection network, using the prepared pedestrian The training dataset is used for final network training. The beneficial effects of the present invention are: the detection method of the present invention uses the residual network as a feature extraction network, and splices it with a multi-scale feature map pedestrian detection network, which can detect complex scenes and large pedestrian scale changes, and is accurate. High rate, high detection efficiency and other advantages.
Description
technical field [0001] The invention relates to the technical field of image recognition, in particular to a pedestrian detection method based on neural network multi-scale feature maps. Background technique [0002] With the development of the economy and the advancement of science and technology, people increasingly hope that computers will be intelligent so that they can deal with problems more effectively and accurately than humans. In the field of computer intelligence, the intelligence of computer vision is a very important part. Nowadays, applications such as automatic driving of cars, unmanned stores, and robot automation are closely related to the intelligence of computer vision. Intelligent target detection technology is one of the core problems that these applications need to solve. At present, the traditional pedestrian detection method trains the classifier by extracting features such as HOG, LBP, and Haar. The obtained model has a good detection rate for pedes...
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): G06K9/62G06K9/00G06N3/04
CPCG06N3/04G06V20/42G06F18/2148
Inventor 舒泓新蔡晓东陈昀王秀英贺光明
Owner CHINACCS INFORMATION IND