What it does This repository is a community-maintained index of synthetic datasets, 3D model repositories, rendering tools, and research papers for computer vision. Think of it as a card catalog for fake data: SunCG, SURREAL, Virtual KITTI, CARLA, AirSim, and dozens more, each tagged and cross-referenced with PDFs, code, and project pages.

The interesting bit The README itself is the product. It uses Google Scholar citekeys as div IDs so references can be hyperlinked bidirectionally—a neat hack that turns a markdown file into a poor man’s citation graph. The maintainers explicitly invite pull requests to keep it current, which is either admirably open or a quiet admission that the field moves faster than any one person can track.

Key highlights

• Covers datasets (Synthia, SceneFlow, Playing for Benchmarks), 3D repositories (ShapeNet), and engines (UnrealCV, UETorch, VizDoom)
• Each entry links paper, code, and project page where available
• Keyword tags (“synthetic human”, “domain”, “rl”) make the list filterable by technique
• Includes workshop and challenge listings (ECCV 2016 VARVAI, CVPR 2017 THOR)
• ~1K stars suggests it has become a de facto starting point for researchers entering the space

Caveats

• No code in this repo itself; it’s purely a curated list
• Coverage thins after 2017; the 2020 section has exactly one entry
• Some links will rot, and the “Total=N” year headers are already optimistic given truncation

Verdict Grab this if you’re starting a project that needs synthetic data and want to avoid reinventing the render pipeline. Skip it if you’re looking for runnable tools—this is the bibliography, not the lab.