Implementation of the paper "Video Frame Interpolation by Plug-and-Play Deep Locally Temporal Embedding"
neuralnet (pip install git+git://github.com/justanhduc/[email protected])
To reproduce the results in the paper, first pickle the UCF data into pkl files. A sample of the pkl file is provided in data. Then the path to a pkl file in DeepLTE_UCF.config. Feel free to change any parameters in the config file. After that, execute
python interpolate_UCF.py DeepLTE_UCF.config (--gpu 0)
To interpolate frames for a video sequence, first extract all frames into a folder. Then specify the path in DeepLTE.config. Feel free to change any parameters in the config file. After that, execute
python interpolate.py DeepLTE.config (--gpu 0)
NOTE: The mosquito noise is generated by the imwrite function of Matlab.
UCF results
Some results from DAVIS
Some results from daily-life videos
MIT license