PyTorch implementation of "Super SloMo: High Quality Estimation of Multiple Intermediate Frames for Video Interpolation" by Jiang et al. [Project] [Paper]
Results on UCF101 dataset using the evaluation script provided by author. The get_results_bug_fixed.sh script was used. It uses motions masks when calculating PSNR, SSIM and IE.
| Method | PSNR | SSIM | IE |
|---|---|---|---|
| DVF | 29.37 | 0.861 | 16.37 |
| SepConv - L_1 | 30.18 | 0.875 | 15.54 |
| SepConv - L_F | 30.03 | 0.869 | 15.78 |
| SuperSloMo_Adobe240fps | 29.80 | 0.870 | 15.68 |
| pretrained mine | 29.77 | 0.874 | 15.58 |
| SuperSloMo | 30.22 | 0.880 | 15.18 |
This codebase was developed and tested with pytorch 0.4.1 and CUDA 9.2.
In order to train the model using the provided code, the data needs to be formatted in a certain manner.
The create_dataset.py script uses ffmpeg to extract frames from videos.
For adobe240fps, download the dataset, unzip it and then run the following command
python data\create_dataset.py --ffmpeg_dir path\to\ffmpeg --videos_folder path\to\adobe240fps\videoFolder --dataset_folder path\to\dataset --dataset adobe240fpsMore Info TBA
You can download the pretrained model trained on adobe240fps dataset here.
More info TBA
| Task | Status |
|---|---|
| Add evaluation script for UCF dataset | TBD |
| Add pretrained model | In Progress |
| Add getting started guide | TBD |
| Add video converter script | In progress |