NEVC - Neural Efficient Video Coding

NEVC is a neural video coding framework designed for highly efficient video compression. By integrating cutting-edge neural network models, NEVC pushes the boundaries of encoding performance and efficiency.

This repository provides access to code, pretrained models, and research papers for various versions of NEVC.

---

📰 Release Notes

NEVC-1.0 - First Release

• Release Date: Sep 5th, 2025
• Key Features:
  • This version implements the core concepts and methods from the paper "EHVC: Efficient Hierarchical Reference and Quality Structure for Neural Video Coding", accepted at ACM MM 2025.

---

📝 Codec Versions Overview

The following table provides details on different versions of the NEVC codec, along with associated papers, code, and checkpoints.

Codec	Paper	Code	Checkpoint
NEVC-1.0 (EHVC)	EHVC paper	EHVC code	EHVC checkpoint

---

🎯 Project Goals

• Develop a neural-based video codec that offers efficient video compression.
• Achieve major enhancements in encoding performance and compression efficiency.

📖 Citation

If you find NEVC or any part of this repository helpful in your research or projects, we kindly ask you to consider citing the following papers:

• EHVC: Efficient Hierarchical Reference and Quality Structure for Neural Video Coding
  Junqi Liao, Yaojun Wu, Chaoyi Lin, Zhipin Deng, Li Li, Dong Liu, Xiaoyan Sun, ACM MM 2025.

  @inproceedings{liao2025ehvc,
    title={EHVC: Efficient Hierarchical Reference and Quality Structure for Neural Video Coding},
    author={Liao, Junqi and Wu, Yaojun and Lin, Chaoyi and Deng, Zhipin and Li, Li and Liu, Dong and Sun, Xiaoyan},
    booktitle={Proceedings of the 33rd ACM International Conference on Multimedia},
    year={2025}
  }
  

📜 License

NEVC is licensed under the BSD 3-Clause Clear License