This repository is a curated list of recent advancements in human avatar technology, covering generation, reconstruction, and editing. It serves researchers and practitioners in computer vision and graphics seeking to stay updated on state-of-the-art techniques for creating and manipulating digital human representations.

How It Works

The list categorizes papers across key areas like Avatar Generation, Per-subject Avatar Reconstruction, Generalizable Avatars, Text-to-Avatar synthesis, Avatar Interaction, Motion Generation, and SMPL Estimation. It highlights recent papers, often referencing their publication venue and year, providing a structured overview of the field's progress.

Highlighted Details

• Extensive coverage of Neural Radiance Fields (NeRF) and Gaussian Splatting techniques for human rendering and reconstruction.
• Numerous papers focus on generating animatable and relightable avatars from single images or monocular video.
• A dedicated section for Text-to-Avatar methods showcases the integration of diffusion models and transformers for text-guided avatar creation.
• Includes papers on 4D human capture, interaction, and motion generation, alongside SMPL estimation methods.

Maintenance and Community

The project is maintained by contributors from ReLER Lab, CCAI, Zhejiang University. Users are encouraged to open issues or pull requests for missed papers.

Licensing and Compatibility

The repository itself is a list of papers and does not contain code. Licensing information for individual papers would need to be checked separately.

Limitations and Caveats

This is a curated list and does not provide code implementations or benchmarks for the listed papers. Users must refer to the original publications for details on methodology, performance, and usage.