Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields

We present an approach to efficiently detect the 2D pose of multiple people in an image. The approach uses a nonparametric representation, which we refer to as Part Affinity Fields (PAFs), to learn to associate body parts with individuals in the image... (read more)

PDF Abstract CVPR 2017 PDF CVPR 2017 Abstract

Datasets


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Multi-Person Pose Estimation COCO CMU-Pose AP 0.618 # 9
Keypoint Detection COCO Part Affinity Fields Validation AP 60.5 # 9
Keypoint Detection COCO test-dev CMU Pose APL 68.2 # 15
APM 57.1 # 14
AP50 84.9 # 12
AP75 67.5 # 12
AR 66.5 # 10
AR50 87.2 # 7
AR75 71.8 # 7
ARL 74.6 # 7
ARM 60.6 # 7
Pose Estimation COCO test-dev CMU-Pose AP 61.8 # 16
AP50 84.9 # 16
AP75 67.5 # 17
APL 68.2 # 16
AR 66.5 # 13
Multi-Person Pose Estimation COCO test-dev CMU-Pose AP 61.8 # 11
APL 68.2 # 8
APM 57.1 # 9
AP50 84.9 # 7
AP75 67.5 # 8
Keypoint Detection MPII Multi-Person Part Affinity Fields mAP@0.5 75.6% # 6
Multi-Person Pose Estimation MPII Multi-Person Part Affinity Fields AP 75.6% # 6

Methods used in the Paper


METHOD TYPE
PAFs
Output Functions