Selective Complementary Features for Multi-Person Pose Estimation
This paper presented a novel approach using Attention Refinement Residual Bottleneck and Information Complement Module for multi-person pose estimation. Challenges, experiments, and results from the COCO dataset evaluation were discussed.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.
E N D
Presentation Transcript
ICIP 2020 IEEE International Conference on Image Processing SELECTIVE COMPLEMENTARY FEATURES SELECTIVE COMPLEMENTARY FEATURES FOR MULTI FOR MULTI- -PERSON POSE ESTIMATION PERSON POSE ESTIMATION Buwei Buwei Li Li, Kai Liu, Yi Ji, Jianyu Yang and Chunping Liu School of Computer Science and Technology, Soochow University
O Outline utline Multi Multi- -Person Pose Estimation Person Pose Estimation Definition and Challenges Our Approach Our Approach Attention Refinement Residual Bottleneck Information Complement Module Experiments Experiments Ablation Study Comparison with Existing Methods Some Qualitative Results Conclusions Conclusions
Multi Multi- -Person Pose Estimation Person Pose Estimation Definition: Definition: Localize the human keypoints for all persons in a natural image target
Multi Multi- -Person Pose Estimation Person Pose Estimation Challenges: Challenges: Complex Background, Self-Occlusion and Occlusion, etc
Our Approach Our Approach Attention Refinement Residual Bottleneck (ARRB) Attention Refinement Residual Bottleneck (ARRB)
Our Approach Our Approach Information Complement Module (ICM) Information Complement Module (ICM)
Experiments Experiments COCO dataset COCO dataset 200K images and 250K person instances labeled with 17 keypoints Evaluation Metrics Evaluation Metrics AP, AP at OKS=0.50:0.05:0.95 (primary challenge metric) AP.50, AP at OKS=0.50 AP.75, AP at OKS=0.75 AP?, AP for medium objects: 322< ???? < 962 APL, AP for large objects: ???? > 962 AR, AR at OKS=0.50:0.05:0.95
Experiments Experiments Ablation Study Ablation Study
Experiments Experiments Ablation Study Ablation Study
Experiments Experiments Ablation Study Ablation Study
Experiments Experiments Comparison Comparison with Existing Methods with Existing Methods
Experiments Experiments Qualitative Qualitative Results Results Self-Occlusion Case Occlusion Case
Conclusions Conclusions Exploit the trade-off between the low-level and high-level feature maps Propose a novel and effective approach with Information Complement Module and Attention Refinement Residual Bottleneck for multi-person pose estimation
