CVPR 2026 Papers with Code: a crowdsourced paper index
/CVPR2026-Papers-with-Code
CVPR 2026 Papers with Code is worth bookmarking if you track computer vision research paper-by-paper rather than through arXiv search. It's early: CVPR 2026 decisions just posted on OpenReview, so most category headers — Agent, GAN, NeRF, Object Detection — sit empty, with only a few dozen papers linked to code so far. Check back through the review cycle rather than expecting a finished list today.
What is CVPR 2026 Papers with Code?
CVPR 2026 Papers with Code is a GitHub repository from amusi that lists CVPR 2026 papers alongside their arXiv links, project pages, and open-source code when authors release it. Entries sit under dozens of topic headers — 3D Gaussian Splatting, vision-language models, medical image segmentation, autonomous driving, and more — filled in as new papers surface. Many headers are still empty since CVPR 2026 decisions just came out.
Research categories covered in the collection
- ✓3D and scene understanding: 3D Gaussian Splatting, 3D point clouds, 3D reconstruction, and 3D human pose estimation each get their own header, with a few entries already linked to arXiv and GitHub code.
- ✓Multimodal and language-vision work: separate headers for MLLM (multimodal large language models), Vision-Language, CLIP, and Prompt-based methods.
- ✓Applied domains get their own space too: medical image segmentation, autonomous driving, and remote sensing each have a dedicated header instead of being buried under a generic 'applications' tag.
- ✓Generation and editing tasks — image generation, video generation, image editing, video editing, and 3D generation — sit in separate headers so they don't blend into detection and segmentation topics.
- ✓Classic CV tasks are pre-created as headers even before any paper fills them: object detection, object tracking, anomaly detection, super-resolution, denoising, depth estimation, stereo matching.
- ✓Each filled-in entry follows the same template — paper title, arXiv link, and a project page or code repository when one exists — which keeps the format skimmable once you know the pattern.
How to use this paper collection for your research
- •Skim category headers for your subfield (say, 3D Reconstruction or Vision-Language) instead of scrolling a flat list of every CVPR 2026 paper.
- •Open the arXiv link first, since many entries don't have a code link yet — check back later or watch the repo for updates.
- •Cross-reference with amusi's sister repos for ICCV 2025 and ECCV 2024 (linked in the README) if you're tracking a topic across multiple conferences, not just CVPR.
- •Open an issue to add a paper you know about — the maintainer explicitly asks readers to submit CVPR 2026 papers and code links themselves.
Strengths
- ✓Actually organized: dozens of specific topic headers instead of one long undifferentiated feed of paper titles.
- ✓Every filled-in entry links straight to the arXiv abstract, so you're one click from the actual paper, not a summary of it.
- ✓Open to contribution — the README explicitly invites readers to submit papers and code links via GitHub issues, so it grows with input beyond just the maintainer.
- ✓Cross-references other conferences: links to the maintainer's ICCV 2025 and ECCV 2024 equivalents if your search spans more than one venue.
What this collection doesn't cover
- △Most category headers are still empty — Agent, Backbone, CLIP, Mamba, GAN, OCR, NeRF, DETR, LLM, NAS, Diffusion Models, Object Detection, and many others have no papers listed yet.
- △Coverage depends entirely on community submissions via GitHub issues, so it can lag well behind what's actually on arXiv or OpenReview.
- △A lot of entries have a paper link but no code link — some are explicitly marked 'Code: None,' meaning the authors haven't released anything to point to.
- △There's no search, filter, or tagging beyond the markdown headers — finding a specific paper means scanning the page or using your browser's find function.
- △License and primary language aren't listed on the repo, so if you plan to reuse or redistribute anything beyond linking to it, you'd need to check with the individual paper authors.
Other ways to find CVPR 2026 papers and code
Frequently asked questions
CVPR 2026 Papers with Code doesn't publish a completion date — the repository fills in as the maintainer and readers submit papers via GitHub issues, and most category headers are still empty since decisions only recently posted on OpenReview.
The README asks readers to open a GitHub issue on the CVPR2026-Papers-with-Code repository to share a CVPR 2026 paper and its code link, which the maintainer then adds to the matching category header.
Not directly — CVPR 2026 Papers with Code is scoped to CVPR 2026 only. The README links out to the same maintainer's separate ICCV 2025 and ECCV 2024 repositories, and to a broader daily-paper-computer-vision roundup, for other conferences and years.
The README doesn't document an update schedule for this specific repository. Entries get added as the maintainer and readers submit them through GitHub issues, so some papers show a paper link with no code, or 'Code: None,' until an implementation is released or reported.
CVPR 2026 accepted 4090 of 16092 submissions, a 25.42% acceptance rate, according to the decisions posted on OpenReview and cited in this repository's README.
CVPR 2026 Papers with Code does have dedicated headers for 3D Gaussian Splatting (3DGS) and for multimodal large language models (MLLM), each already listing a few papers with arXiv links and, in some cases, GitHub code.
Who should try it — and who should skip
CV researchers and grad students tracking one CVPR 2026 subfield — say, 3D Gaussian Splatting or vision-language models — get real value from checking back on this repository as entries fill in through the review cycle. Skip it for now if you need full coverage. Most headers are still empty because CVPR 2026 decisions only just posted on OpenReview, so a direct arXiv or OpenReview search will surface more papers today than this collection will.
Related repositories
Still deciding about CVPR2026-Papers-with-Code?
One click hands the question to an AI along with this page — see what it says about CVPR2026-Papers-with-Code.