Cascade Transformer Decoder based Occluded Pedestrian Detection with Dynamic Deformable Convolution and Gaussian Projection Channel Attention Mechanism - Details

Author：

Ma, C. (Ma, C..) | Zhuo, L. (Zhuo, L..) | Li, J. (Li, J..) | Zhang, Y. (Zhang, Y..) | Zhang, J. (Zhang, J..)

Indexed by：

EI Scopus SCIE

Abstract：

Occluded　pedestrian　detection　is　very　challenging　in　computer　vision,　because　the　pedestrians　are　frequently　occluded　by　various　obstacles　or　persons,　especially　in　crowded　scenarios.　In　this　paper,　an　occluded　pedestrian　detection　method　is　proposed　under　a　basic　DEtection　TRansformer　(DETR)　framework.　Firstly,　Dynamic　Deformable　Convolution　(DyDC)　and　Gaussian　Projection　Channel　Attention　(GPCA)　mechanism　are　proposed　and　embedded　into　the　low　layer　and　high　layer　of　ResNet50　respectively,　to　improve　the　representation　capability　of　features.　Secondly,　Cascade　Transformer　Decoder　(CTD)　is　proposed,　which　aims　to　generate　high-score　queries,　avoiding　the　influence　of　low-score　queries　in　the　decoder　stage,　further　improving　the　detection　accuracy.　The　proposed　method　is　verified　on　three　challenging　datasets,　namely　CrowdHuman,　WiderPerson,　and　TJU-DHD-pedestrian.　The　experimental　results　show　that,　compared　with　the　state-of-the-art　methods,　it　can　obtain　a　superior　detection　performance.　IEEE

Keyword：

Feature extraction Decoding Task analysis Cascade Transformer Decoder Transformers Kernel Gaussian project channel attention mechanism Convolution Object detection Occluded pedestrian detection Dynamic deformable convolution

Author Community：

[ 1 ] [Ma C.]Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing, China
[ 2 ] [Zhuo L.]Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing, China
[ 3 ] [Li J.]Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing, China
[ 4 ] [Zhang Y.]Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing, China
[ 5 ] [Zhang J.]Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing, China

Reprint Author's Address：

Email：

Show more details

Related Keywords：

SCDA-Net: Structure Completion and Density Awareness Network for LiDAR-Based 3D Object Detection
2025，IEEE ROBOTICS AND AUTOMATION LETTERS
Spectral-Spatial Large Kernel Attention Network for Hyperspectral Image Classification
2024，IEEE Transactions on Geoscience and Remote Sensing
Cascaded Segmented Matting Network for Human Matting
2021，IEEE ACCESS
Hierarchical Graph Convolutional Networks for Structured Long Document Classification
2022，IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS

Source ：

IEEE Transactions on Multimedia

ISSN： 1520-9210

Year： 2023

Volume： 25

Page： 1-8

7 . 3 0 0

JCR@2022

ESI Discipline： COMPUTER SCIENCE;

ESI HC Threshold：19

Cited Count：

WoS CC Cited Count： 0

SCOPUS Cited Count： 20

ESI Highly Cited Papers on the List： 0 Unfold All

WanFang Cited Count：

Chinese Cited Count：

30 Days PV： 24

Affiliated Colleges：

Get Fulltext

DOI Library Discovery Baidu Scholar Search SCOPUS

Type
Departments

All Years Choose Year From to