A living index of CVPR 2026 papers, sorted before the ink dries

What it does

This is not software — it’s a curated list, in the tradition of awesome-repos. It tracks CVPR 2026 papers and links each to its arXiv page, code repository, and project site where they exist. Entries are grouped into ~80 categories spanning the usual suspects: 3DGS, NeRF, diffusion, MLLMs, segmentation, autonomous driving, medical imaging, and more. The README notes a 25.42% acceptance rate (4090 of 16092 submissions), for those keeping score at home.

The interesting bit

The taxonomy itself is a snapshot of what computer vision currently is — “Embodied AI,” “Spatial Intelligence,” and “MLLM” get their own sections now, while many classic categories sit empty. It’s a decent proxy for where the field’s center of gravity has moved.

Key highlights

• Papers are filed under ~80 fine-grained topics, from Gaussian Splatting to video compression.
• Each entry carries paper, code, and project links when available — some papers list code, some only project pages.
• Community-driven: the maintainer explicitly invites issue submissions of new papers and repos.
• Part of a family of sibling repos covering ICCV 2025 and ECCV 2024, plus a daily CV paper digest.

Caveats

• Many sections are still empty placeholders — the list is early and filling in as authors release code.
• The README is primarily in Chinese, and it promotes a paid community (“CVer 知识星球”) alongside the free content.
• It’s a link index, not a tool: no code, no automation, just markdown.

Verdict

Worth a bookmark if you follow CV research and want a one-stop index of CVPR 2026 with code links. Skip it if you expected anything runnable — the value here is curation, not software.