MovingColor: Seamless Fusion of Fine-grained Video Color Enhancement

Yi Dong1, Yuxi Wang1, Zheng Fang1, Wenqi Ouyan1, Xianhui Lin2, Zhiqi Shen1*, Peiran Ren2*, Xuansong Xie2 Qingming Huang3
1Nanyang Technological University, Singapore 2Institute for Intelligent Computing, Alibaba Group, China 3University of Chinese Academy of Sciences, China *Corresponding Author
Paper Supplementary Materials ArXiv Video Code
Visual 1

MovingColor tackles the problem of natural fusion of fine-grained video color adjustments. It can achieve refined fusion of regional color adjustments and fuse the edit seamlessly into the input video and keep the temporal consistency in the meantime.

Video Illustration

Comparison with State-of-the-art Methods

Architecture

The results show that MovingColor is the only method that can achieve both spatial temporal consistency while with minimal color deviation in the non-edge area.

D5 Dataset

Architecture

Dataset release pending copyright review.

BibTeX

@inproceedings{dong2024movingcolor,
      author = {Dong, Yi and Wang, Yuxi and Fang, Zheng and Ouyang, Wenqi and Lin, Xianhui and Shen, Zhiqi and Ren, Peiran and Xie, Xuansong and Huang, Qingming},
      title = {MovingColor: Seamless Fusion of Fine-grained Video Color Enhancement},
      year = {2024},
      isbn = {9798400706868},
      publisher = {Association for Computing Machinery},
      address = {New York, NY, USA},
      url = {https://doi.org/10.1145/3664647.3681130},
      doi = {10.1145/3664647.3681130},
      booktitle = {Proceedings of the 32nd ACM International Conference on Multimedia},
      pages = {7454–7463},
      numpages = {10},
      keywords = {color fusion, video color enhancement, video editing},
      location = {Melbourne VIC, Australia},
      series = {MM '24}
      }