Vehicle detection method based on key point heat map

A technology of vehicle detection and key points, applied in the field of image processing, can solve the problems of accuracy gap, slow down the speed of training, multiple anchor frames, etc.

Pending Publication Date: 2019-11-22
CHINA JILIANG UNIV
View PDF1 Cites 20 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this kind of method also has certain defects. During the training process, more anchor boxes will be generated, which will cause the imbalance of positive and negative samples and slow down the speed of training.
Secondly, the setting of the anchor box will introduce more hyperparameters to the network, and the s...

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Vehicle detection method based on key point heat map
  • Vehicle detection method based on key point heat map
  • Vehicle detection method based on key point heat map

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0027] The specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0028] The vehicle detection method based on key point thermal map of the present invention, comprises the following steps:

[0029] Step 1: Training and test data preparation: download and obtain vehicle public data sets KITTI and UA-DETRAC, use web crawler algorithm to obtain vehicle pictures from the Internet, use open source vehicle detection algorithm to extract vehicle images from road monitoring video data, and obtain a total of samples M open;

[0030] Step 2: Data annotation: such as figure 1 Mark four key points on the M pieces of vehicle data described in step 1, and divide the vehicles into four categories according to the position of the vehicle body displayed in the picture (four directions of front, rear, left, and right) when marking. The key points of the two types of vehicles in the front and rear directions are the upp...

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

No PUM Login to view more

Abstract

The invention discloses a vehicle detection method based on a key point heat map, and aims to provide a novel vehicle detection method by combining specific characteristics of a vehicle with prediction of the key point heat map, and the technical key point is as follows: (1) labeling a key point data of a specific attribute position of the vehicle; (2) extracting features of the image of the vehicle by adopting a funnel network; and (3) adopting improved key point pooling to obtain characteristics composed of a key point heat map, an embedded vector and an offset vector, supplementing global information of the positioned vehicle by utilizing the characteristics of the positioned license plate or wheel, removing an incorrect target frame, finally obtaining a target positioning frame with relatively good robustness, and predicting the direction of the vehicle.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a vehicle detection and vehicle direction recognition method based on a key point heat map. Background technique [0002] With the rapid development of deep learning and artificial intelligence, vehicle detection related technologies are gradually being applied and developed in intelligent transportation and smart communities, helping users obtain route traffic flow, saving time and ensuring safety. As a meticulous application of a specific category of object detection technology, vehicle detection technology has a unique feature representation. One of the most popular methods at present is the detection method of setting the anchor frame, by pre-setting the anchor frame with different width and height ratios to predict the candidate frame, so as to obtain a higher detection accuracy. However, this kind of method also has certain defects. During the training process, ...

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
no application Login to view more
IPC IPC(8): G06K9/62G06F16/951G06F16/58
CPCG06F16/951G06F16/5866G06V2201/08G06F18/241G06F18/214
Inventor 章东平葛俊
Owner CHINA JILIANG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap