ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation

Abstract

Although no specific domain knowledge is considered in the design, plain vision transformers have shown excellent performance in visual recognition tasks. However, little effort has been made to reveal the potential of such simple structures for pose estimation tasks. In this paper, we show the surprisingly good capabilities of plain vision transformers for pose estimation from various aspects, namely simplicity in model structure, scalability in model size, flexibility in training paradigm, and transferability of knowledge between models, through a simple baseline model called ViTPose. Specifically, ViTPose employs plain and non-hierarchical vision transformers as backbones to extract features for a given person instance and a lightweight decoder for pose estimation. It can be scaled up from 100M to 1B parameters by taking the advantages of the scalable model capacity and high parallelism of transformers, setting a new Pareto front between throughput and performance. Besides, ViTPose is very flexible regarding the attention type, input resolution, pre-training and finetuning strategy, as well as dealing with multiple pose tasks. We also empirically demonstrate that the knowledge of large ViTPose models can be easily transferred to small ones via a simple knowledge token. Experimental results show that our basic ViTPose model outperforms representative methods on the challenging MS COCO Keypoint Detection benchmark, while the largest model sets a new state-of-the-art. The code and models are available at https://github.com/ViTAE-Transformer/ViTPose.

Neurips 2022. 81.1 mAP on MS COCO Keypoint Detection test-dev set. V2: Update Multi-task training results: 92.8 AP on OCHuman, 78.3 AP on CrowdPose, 94.3 PCKh on MPII, and 43.2 AP on AI Challenger

Cited by in corpus (17)

• Vision-Based Human Pose Estimation via Deep Learning: A Survey
• SHaRPose: Sparse High-Resolution Representation for Human Pose Estimation
• CLAMP: Prompt-based Contrastive Learning for Connecting Language and Animal Pose
• Neuromorphic High-Frequency 3D Dancing Pose Estimation in Dynamic Environment
• HRPVT: High-Resolution Pyramid Vision Transformer for medium and small-scale human pose estimation
• Under the Cover Infant Pose Estimation using Multimodal Data
• Pose2Trajectory: Using Transformers on Body Pose to Predict Tennis Player's Trajectory
• SUPER: Seated Upper Body Pose Estimation using mmWave Radars
• Human Pose Estimation from Ambiguous Pressure Recordings with Spatio-temporal Masked Transformers
• Improving 2D Human Pose Estimation in Rare Camera Views with Synthetic Data
• Towards High Performance One-Stage Human Pose Estimation
• SniffyArt: The Dataset of Smelling Persons
• ARTS: Semi-Analytical Regressor using Disentangled Skeletal Representations for Human Mesh Recovery from Videos
• BOTH2Hands: Inferring 3D Hands from Both Text Prompts and Body Dynamics
• Precision at Scale: Domain-Specific Datasets On-Demand
• The Influence of Faulty Labels in Data Sets on Human Pose Estimation
• TrackID3x3: A Dataset and Algorithm for Multi-Player Tracking with Identification and Pose Estimation in 3x3 Basketball Full-court Videos