xinghaochen/awesome-hand-pose-estimation

What this project is

awesome-hand-pose-estimation is a curated reference list (an "awesome list") of research resources about hand pose estimation — the computer-vision task of figuring out where a hand is and how its joints are positioned, usually from images or depth data.

It is not software you run. It is a reading list that collects papers, datasets, and other materials, each with links to PDFs and sometimes to code or project pages.

What it contains

Based on the README, the list is organized into these sections:

• Evaluation — points to a folder called evaluation with details on how hand-pose performance is measured.
• arXiv Papers — pre-print papers, each with title, authors, and links to the arXiv page and PDF. Some also link to code, project pages, or datasets.
• Journal Papers — peer-reviewed journal articles, split into two groups:
  • TPAMI / IJCV (two top computer-vision journals)
  • Other Journals (e.g., TIP, TCSVT, Neurocomputing, Sensors, etc.)
• Conference Papers — papers from major conferences (CVPR, ECCV, ICCV, and others), organized by year from 2014 to 2025. (The README's table of contents lists this section, but the detailed entries are not shown in the provided text.)
• Theses — academic theses (listed in the table of contents, details not shown).
• Datasets — grouped into Depth, RGB+Depth, and RGB datasets (listed in the table of contents, details not shown).
• Workshops and Challenges — related events (listed in the table of contents, details not shown).
• Other Related Papers — additional papers (listed in the table of contents, details not shown).

How entries are formatted

Each paper entry typically includes:

• The title
• The authors (with * marking equal contribution)
• A link to the PDF (arXiv, IEEE, ScienceDirect, etc.)
• Sometimes extra links: code, project page, or dataset

Example entries (from the README)

• TriHorn-Net — a depth-based 3D hand pose estimation model, with code on GitHub.
• FrankMocap — fast monocular 3D hand and body motion capture, with a project page.
• HandAugment — a data augmentation method for depth-based 3D hand pose estimation, with code.
• A Survey on 3D Hand Pose Estimation: Cameras, Methods, and Datasets — a survey paper.

How to use it

The README invites contributions ("Feel free to contribute!"). The list is meant to be browsed — you can jump to a section via the table of contents, find papers by year or venue, and follow links to read or download them.

Key takeaway

This is a maintained bibliography for researchers and developers working on hand pose estimation. It organizes the field's literature by source (arXiv, journals, conferences), by year, and by topic (datasets, challenges, etc.), making it easy to find relevant work.