A curated map of ECCV 2024 work you can actually compile

What it does This repository is a manually curated index of ECCV 2024 papers that ship open-source code. It collects arXiv links, project pages, and GitHub repositories across dozens of computer vision subfields—from 3D Gaussian Splatting and diffusion models to medical image segmentation and autonomous driving—so researchers can browse by task instead of hunting through proceedings. Entries are added via community submissions, and the repo links to parallel collections for CVPR 2024 and earlier ECCV years.

The interesting bit The breadth is the product: it spans 50-odd categories, including niche areas like mmWave radar scene flow and fractal feature maps for tubular structure segmentation, acting as a human-filtered signal amid conference noise. It is essentially glue code and curation, but the editorial sorting saves you from parsing hundreds of PDF abstracts to find reproducible work.

Key highlights

• Covers 50+ CV topics, including 3DGS, NeRF, Mamba, diffusion models, medical imaging, and autonomous driving
• Direct links to paper PDFs, project pages, and source code for each entry
• Community-maintained via GitHub issues; explicitly invites submissions for missing papers
• Companion repositories exist for CVPR 2024, ECCV 2022, and ECCV 2020
• Roughly 2,300 stars suggest it has become a popular reference point

Caveats

• Many topic sections (e.g., Avatars, Backbone, CLIP, GAN, LLM) are currently empty placeholders with no entries
• Coverage is uneven; some categories contain only one or two papers while others are more populated
• Primarily documented in Chinese, though paper titles and links are in English

Verdict Worth bookmarking if you’re mining ECCV 2024 for reproducible baselines or tracking trends across computer vision tasks. Skip it if you expect automated search, paper abstracts, or systematic quality filtering—this is a link directory, not a review journal.