Practical overview
When it fits
Use it when you need a self-hosted PyTorch starting point for image-based human-pose estimation or for training experiments with ViTPose variants.
Who it is for
- Developers who want to run human-pose estimation on images using a PyTorch codebase.
- Researchers or practitioners exploring training or fine-tuning ViTPose variants.
Documented uses
- Run the supplied inference script on an image.
- Invoke training with a selected model variant and configuration file.
- Compare Base, Large and Huge configurations for a pose-estimation task.
Requirements & limits
The repository is public and licensed under Apache-2.0.
- Requires Python 3.8 or later and PyTorch 1.9.0 or later, plus torchvision, NumPy, OpenCV, Pillow and tqdm.
- The three listed configurations all use a 256×192 input size and have separate configuration files and pretrained weights.
- The README describes the implementation as a simple version without MMCV.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Run pose estimation on an image
Set up the repository and run its documented inference script on an image.
Before you start
- Python 3.8 or later
- PyTorch 1.9.0 or later
- torchvision, NumPy, opencv-python, Pillow and tqdm
Steps
Clone the repository and enter its directory.
Install dependencies with the command below.
pip install -r requirements.txtRun the command below, replacing the example path with the image to process.
python inference.py --image-path './examples/img1.jpg'
Expected outcome
The repository's inference script is invoked with the selected image path for human-pose estimation.
Guide limitations
- The README does not specify output-file behavior or inference options beyond the image-path argument.
Start a ViTPose training run
Set up the repository and invoke its training script with a configuration and model variant.
Before you start
- Python 3.8 or later
- PyTorch 1.9.0 or later
- torchvision, NumPy, opencv-python, Pillow and tqdm
- A configuration file path to pass to the training script
Steps
Clone the repository and enter its directory.
Install dependencies with the command below.
pip install -r requirements.txtRun the command below, substituting the intended configuration path and using
b,lorhfor Base, Large or Huge.python train.py --config-path config.yaml --model-name 'b'
Expected outcome
The training script is invoked for the selected ViTPose variant and configuration.
Guide limitations
- The README does not describe dataset preparation, configuration requirements or training outputs.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 15 October 2026
- Original website title
- jaehyunnn/ViTPose_pytorch · Official homepage title · retained in its source language