图像颜色处理:颜色恒常性算法库 (Anole)

发布时间:2022-08-15 打印[ ]

        Anole颜色恒常性算法库。学术上的颜色恒常性,在工业界≈AWB自动白平衡+CCM颜色矫正矩阵。但是这个领域没有一个规范化、公平、可比较的平台,在RAW数据处理方面缺乏接口、压缩域的数据因为丢失了光照线性变换关系而无法复用,学术界很多sota算法都只有论文没有代码共享,为了解决上述问题,我们做了大量的工作,复现了大量代码,围绕我们自研的核心代码,按照六个类别组织了这个算法库。据我们所知,这也是全球首个颜色恒常性平台级算法库。

       开源项目链接:https://github.com/mRobotit/Anole

image.png


相关论文:

  • [1] Chunxiao Li, Xuejing Kang, Anlong Ming*WBFlow: Few-shot White Balance for sRGB Images via Reversible Neural Flows, in Proceedings of the 32st International Joint Conference on Artificial Intelligence (IJCAI), 2023.  

    代码https://github.com/mRobotit/WBFlow

  • [2] Chunxiao Li, Xuejing Kang, Zhifeng Zhang, Anlong Ming*, SWBNet: A Stable White Balance Network for sRGB images, the 37th AAAI Conference on Artificial Intelligence (AAAI), 2023. 

    代码:https://github.com/mRobotit/SWBNet

  • [3] Zhifeng Zhang, Xuejing Kang, Anlong Ming*Domain Adversarial Learning for Color Constancy, in Proceedings of the 31st International Joint Conference on Artificial Intelligence and the 25th European Conference on Artificial Intelligence (IJCAI), 2022.  

    代码:https://github.com/mRobotit/DALCC

  • [4] Yuxiang Tang, Xuejing Kang, Chunxiao Li, Zhaowen Lin, Anlong Ming*, Transfer Learning for Color Constancy via Statistic Perspective, the 36th AAAI Conference on Artificial Intelligence (AAAI), 2022. 

    代码:https://github.com/mrobotit/TLCC