Counting heads: a curated research index

What it does

This is a curated index of the crowd-counting research landscape. It catalogs datasets, benchmarks, open-source code, and academic papers—sorting them into categories like transformers, lightweight models, semi-supervised learning, and attention mechanisms. Think of it as a field guide for a narrow but crowded corner of computer vision.

The interesting bit

The maintainers impose a useful discipline: arXiv preprints are listed but deliberately excluded from the performance leaderboard until they are formally published, which keeps the benchmark honest while still tracking the bleeding edge.

Key highlights

• Categorized paper lists spanning surveys, domain adaptation, video crowd counting, and network architecture search
• Curated dataset releases including NWPU-Crowd, JHU-CROWD++, and RGBT-CC
• Links to practical tooling like the C^3 Framework PyTorch codebase and the CCLabeler web annotation tool
• Active tracking of workshops, special issues, and benchmark challenges
• A leaderboard that filters out unpublished arXiv papers from official rankings

Verdict

Worth bookmarking if you are building surveillance analytics, event-safety tools, or just trying to benchmark a new density-estimation model; skip it if you are looking for a single turnkey counting application.