This repository serves as a curated and continuously updated collection of survey papers and foundational research in computer vision, targeting researchers and practitioners seeking to stay abreast of the latest advancements. It provides categorized links to recent survey papers across various sub-fields, facilitating efficient literature review and knowledge discovery.

How It Works

The project functions as a dynamic bibliography, meticulously categorizing and linking to survey papers published in major computer vision conferences and journals. It organizes research by sub-fields such as object detection, autonomous driving, medical imaging, and generative models, with a focus on recent publications (2020-2025). This structured approach allows users to quickly identify and access relevant state-of-the-art reviews.

Quick Start & Requirements

• Access to the curated list of papers is directly available via the README.
• No installation or specific software requirements are needed to browse the collection.
• Links to download paper collections are provided via a WeChat public account.
• Relevant links: 2025-CV-Surveys, CVPR-2025-Papers

Highlighted Details

• Comprehensive coverage of computer vision sub-fields, including emerging areas like multimodal LLMs and diffusion models.
• Focus on recent publications, with dedicated sections for papers from 2020 through 2025.
• Inclusion of links to associated code repositories for many survey papers.
• Regular updates to reflect the rapid pace of research in computer vision.

Maintenance & Community

The project is maintained by "52CV" and provides a WeChat QR code for joining a community discussion group.

Licensing & Compatibility

The repository itself does not host the papers but links to them. The licensing of individual papers would depend on their original publication venues.

Limitations & Caveats

The primary method of accessing paper collections is through a WeChat public account, which may present a barrier for users outside of that ecosystem. The repository is a curated list and does not provide direct access to the full text of the papers.