Video Recoloring via Spatial-Temporal Geometric Palettes
This is the source code of the paper: Video Recoloring via Spatial-Temporal Geometric Palettes, authors: Zheng-Jun Du, Kai-Xiang Lei, Kun Xu, Jianchao Tan, Yotam Gingold. Please fell free to contact us if you have any questions, email: [email protected]
Requirements
Windows 10
Microsoft Visual Studio 2015
OpenCV 4.1
Nlopt 2.4.2 (included in "VideoPaletteExtraction/ThirdParty")
Qhull (included in "VideoPaletteExtraction/ThirdParty")
Python 3.7
Qt 5.14.1(for Video recoloring GUI)
Directories
- data: the video "season" that shown in our paper
- VideoPaletteExtraction: the video palette extraction program
- RecolorGUI: the video recoloring GUI
Usage
-
Generate video frames' convex hulls with Jianchao Tan's method: https://github.com/JianchaoTan/fastLayerDecomposition. Here I provide the frames' convex hulls in "data/season/" , user need not to generate it again.
-
Run the VideoPaletteExtraction program, the generated video palette will save in "data/season/skew_polyhedron_palette"
-
Run "data/scripts/images_to_video_file.py" to generate the original uncompressed video, which will save in "data/season/skew_polyhedron_palette";
-
Run the RecolorGUI to recolor the video:
a) click "Open Video and Palette" to load the video and palette, which locate in "data/season/skew_polyhedron_palette"
b) click "Calculate weights" to calculate the MVC weights
c) change the palette colors in the "Palette Timeline" view
References
[1] Du Z J, Lei K X, Xu K, et al. Video recoloring via spatial-temporal geometric palettes[J]. ACM Transactions on Graphics (TOG), 2021, 40(4): 1-16.
[2] Wang Y, Liu Y, Xu K. An Improved Geometric Approach for Palette‐based Image Decomposition and Recoloring[C]//Computer Graphics Forum. 2019, 38(7): 11-22.
[3] Tan J, Lien J M, Gingold Y. Decomposing images into layers via RGB-space geometry[J]. ACM Transactions on Graphics (TOG), 2016, 36(1): 1-14.
