AID: Pushing the Performance Boundary of Human Pose Estimation with Information Dropping Augmentation

17 Aug 2020  ·  Junjie Huang, Zheng Zhu, Guan Huang, Dalong Du ·

Both appearance cue and constraint cue are vital for human pose estimation. However, there is a tendency in most existing works to overfitting the former and overlook the latter. In this paper, we propose Augmentation by Information Dropping (AID) to verify and tackle this dilemma. Alone with AID as a prerequisite for effectively exploiting its potential, we propose customized training schedules, which are designed by analyzing the pattern of loss and performance in training process from the perspective of information supplying. In experiments, as a model-agnostic approach, AID promotes various state-of-the-art methods in both bottom-up and top-down paradigms with different input sizes, frameworks, backbones, training and testing sets. On popular COCO human pose estimation test set, AID consistently boosts the performance of different configurations by around 0.6 AP in top-down paradigm and up to 1.5 AP in bottom-up paradigm. On more challenging CrowdPose dataset, the improvement is more than 1.5 AP. As AID successfully pushes the performance boundary of human pose estimation problem by considerable margin and sets a new state-of-the-art, we hope AID to be a regular configuration for training human pose estimators. The source code will be publicly available for further research.

PDF Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Multi-Person Pose Estimation COCO minival ResNet50 AP 75.3 # 3
Multi-Person Pose Estimation COCO minival HRNet-W48plus AP 79.1 # 1
Multi-Person Pose Estimation COCO minival HRNet-W32 AP 77.8 # 2
Multi-Person Pose Estimation COCO test-dev HRNet-W48plus AP 78.7 # 2
Multi-Person Pose Estimation COCO test-dev HRNet-W32 AP 76.2 # 3
Multi-Person Pose Estimation COCO test-dev ResNet50 AP 73.7 # 4

Methods