DeepLTE

Implementation of the paper "Video Frame Interpolation by Plug-and-Play Deep Locally Temporal Embedding"

Requirements

Theano

neuralnet (pip install git+git://github.com/justanhduc/neuralnet.git@0.0.5a0)

How to run

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)

Results

NOTE: The mosquito noise is generated by the imwrite function of Matlab.

UCF results

Some results from DAVIS

Some results from daily-life videos

License

MIT license