This repository serves as a curated collection of papers and their associated code for the CVPR 2025 conference, targeting researchers and practitioners in computer vision. It aims to provide a centralized, up-to-date resource for the latest advancements and their implementations, facilitating easier access to cutting-edge research.

How It Works

The repository organizes papers by research topics, ranging from foundational areas like object detection and segmentation to emerging fields such as embodied AI and multimodal large language models. Each entry typically includes a link to the paper's abstract (e.g., on arXiv) and a direct link to the official code repository, enabling users to quickly find and experiment with new techniques.

Quick Start & Requirements

• Installation: Varies per project; typically involves cloning the respective code repository and following its specific setup instructions (e.g., pip install -r requirements.txt).
• Prerequisites: Common dependencies include Python, PyTorch/TensorFlow, and often specific CUDA versions for GPU acceleration. Some projects may require large datasets or pre-trained models.
• Resources: Setup time and resource requirements are project-specific, with many requiring significant GPU memory and processing power for training or inference.
• Links:
  • CVPR 2025 Decisions: OpenReview
  • Previous CVPR/ECCV Papers: amusi/daily-paper-computer-vision

Highlighted Details

• Comprehensive coverage of CVPR 2025 accepted papers.
• Categorization across a wide spectrum of computer vision subfields.
• Direct links to both research papers and their corresponding open-source code implementations.
• Inclusion of projects related to emerging areas like Gaussian Splatting, Mamba-based vision models, and multimodal LLMs.

Maintenance & Community

• The repository is actively maintained by the amusi user.
• Community engagement is encouraged through GitHub Issues for submitting new papers and projects.
• A WeChat group is mentioned for academic exchange.

Licensing & Compatibility

• Licenses are specific to each linked project and are not consolidated here. Users must check the license of each individual repository.
• Compatibility for commercial or closed-source use depends entirely on the licenses of the individual projects.

Limitations & Caveats

This repository is a curated list and does not host the code itself; it acts as a directory. The quality, completeness, and maintenance status of the linked code projects vary. Users must independently verify the functionality and licensing of each external repository.