Unlock instant, AI-driven research and patent intelligence for your innovation.

A vehicle-mounted multi-target recognition tracking prediction method

A prediction method and multi-target technology, applied in the field of multi-target recognition and tracking prediction on the vehicle side, can solve problems such as complexity, difficult to greatly improve the performance of assisted driving technology, and increase the cost of system implementation, and achieve small model size, real-time recognition and tracking and Predictive, conducive to the effect of rapid deployment

Active Publication Date: 2022-05-17
GUILIN UNIV OF ELECTRONIC TECH +1
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Due to the extremely complex actual road conditions, it is difficult to greatly improve the performance of assisted driving technology based on traditional target detection
Existing autonomous driving technologies, although they can rely on advanced radar systems to compensate, greatly increase the cost of system implementation

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
  • A vehicle-mounted multi-target recognition tracking prediction method
  • A vehicle-mounted multi-target recognition tracking prediction method
  • A vehicle-mounted multi-target recognition tracking prediction method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0036] Below in conjunction with specific case the present invention is described further:

[0037] Such as figure 1 As shown, a vehicle-side multi-target recognition tracking prediction method, the method is a vehicle-side multi-target recognition tracking prediction method based on the fusion of YOLOv5s and FairMOT, including the following steps:

[0038] (1) YOLOv5s algorithm multi-target recognition model training in traffic scenarios

[0039] ①The establishment and marking of the traffic scene data set, through the shooting of a large number of pictures in the actual scene and the collection of surveillance video in the road traffic scene, use lablimg to mark the sample image, and use VOTT (Visual ObjectTagging Tool) to mark the video data , it should be pointed out that these two types of marking software are essentially marking a single image. The difference is that the video marking software has functions such as screening sample images and rough automatic marking bas...

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 multi-target recognition and tracking prediction method at the vehicle end, which is based on the fusion of YOLOv5s (You Only Look Once v5s) and FairMOT (Fair Multi-Object Tracking). Deep learning object detection technology quickly and accurately detects vehicles, pedestrians, obstacles, etc. in front of the road, and integrates the YOLOv5s model into the FairMOT architecture detection module to perform target detection, re-identification and tracking in a single network, and realize the position of traffic targets in front of vehicles on the road Detection, type recognition, and multi-target motion trajectory tracking, so as to achieve the prediction of driving behaviors such as changing lanes, following cars, and decelerating traffic targets in front of the vehicle.

Description

technical field [0001] The invention belongs to the field of automatic driving in intelligent transportation. In the case of mixed traffic, the automatic driving vehicle needs to quickly and accurately identify and track the vehicles, pedestrians, obstacles and other targets in front, so as to realize the lane change and car-following of the targets ahead. , deceleration and other driving behavior prediction problems, a vehicle-side multi-target recognition tracking prediction method based on the fusion of YOLOv5s and FairMOT is proposed. Background technique [0002] In recent years, autonomous driving with the aim of greatly reducing accidents has gradually become a social need. The combination of automotive preventive safety technology and object detection algorithms has developed into an autonomous driving assistance technology that can automate part of car driving performed by humans and is expected to further improve the safety of car driving. In autonomous driving te...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06V20/58G06V10/82G06N3/04G06N5/04G06T5/50
CPCG06T5/50G06N5/04G06T2207/20081G06T2207/30252G06V40/103G06V20/56G06N3/045Y02T10/40
Inventor 万千刘华磊彭国庆郑钰谢振友
Owner GUILIN UNIV OF ELECTRONIC TECH