Patent · US Active

Method of segmenting pedestrians in roadside image by using convolutional network fusing features at different scales

US11783594B2 · kind B2 · utility

0Cited by

2References

1Claims

0Family size

Assignee

SOUTHEAST UNIVERSITY · CN

Inventors

Xu Li · Nanjing, CN
Zhiyong Zheng · Tangxia, CN
Kun Wei · Lanzhou, CN

Key dates

Filing date	May 16, 2019
Grant date	Oct 10, 2023
Priority date	—
Expiry date	Mar 16, 2040

Classification

Technology area (CPC Y)Emerging Cross-Sectional Technologies
CPC primaryY02T10/40
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

The present invention discloses a method for segmenting pedestrians in roadside images using a variable-scale multi-feature fusion convolutional network. It addresses the challenge of significant changes in pedestrian scale by using two parallel convolutional neural networks to extract the local and global features at different scales, and then fusing them to obtain a variable-scale multi-feature fusion convolutional neural network, and this network is trained using roadside pedestrian images to realize accurate pedestrian segmentation, avoiding issues with boundary fuzziness and missing segments commonly found in single-network methods.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.