ViTPose (simple version w/o mmcv)

An unofficial implementation of ViTPose [Y. Xu et al., 2022]

Installation

# Clone the repository
git clone https://github.com/jaehyunnn/ViTPose_pytorch.git
cd ViTPose_pytorch

# Install dependencies
pip install -r requirements.txt

Requirements:

• Python >= 3.8
• PyTorch >= 1.9.0
• torchvision
• numpy, opencv-python, pillow, tqdm

Pretrained Models & Configs

Model	Backbone	Input Size	Config	Weights
ViTPose-B	ViT-Base	256x192	config	download
ViTPose-L	ViT-Large	256x192	config	download
ViTPose-H	ViT-Huge	256x192	config	download

Usage

Inference

python inference.py --image-path './examples/img1.jpg'

Training

python train.py --config-path config.yaml --model-name 'b'

• model_name: b (Base), l (Large), h (Huge)

---

Reference

All codes were written with reference to the official ViTPose repo.