DVTSR

PyTorch Implementation of "Robust Temporal Super-Resolution for Dynamic Motion Video"

Related work

PyTorch Implementation of "Densely Connected Hierarchical Network for Image Denoising", second place winner of sRGB track and third place winner of Raw-RGB track on NTIRE 2019 Challenge on Real Image Denoising. (code)

PyTorch reimplementation of "Optical Flow Estimation using a Spatial Pyramid Network" by Simon Niklaus. (code)

Dependencies

Python 3.6
PyTorch 1.0.0
MATLAB
TensorFlow
TensorBoard
TensorBoardX
TorchSummary

Data

We used REDS VTSR dataset for training.
To generate training data, use the matlab code generate_train.m

Training

As an example, use the following command to use our training codes

python main_tsr.py --lr 1e-4 --step 2 --cuda True --train_data0 ./train_data0.h5 --train_data1 ./train_data1.h5 --train_label ./train_label.h5 --valid_data0 ./valid_data0.h5 --valid_data1 ./valid_data1.h5 --valid_label ./valid_label.h5 --gpu 0,1 --checkpoint ./checkpoint

If you want to train t=1/4, change 9th line of the main_tsr.py to from model_tsr_14, import model_tsr
If you want to train t=1/2, change 9th line of the main_tsr.py to from model_tsr_12, import model_tsr
If you want to train t=3/4, change 9th line of the main_tsr.py to from model_tsr_34, import model_tsr
Also, you have to give right label for the target time.
There are other options you can choose.
Please refer to the code.

Test

As an example, use the following command to use our test codes

python test_tsr.py --cuda True --model0 ./model0.pth --data ./path/to/data --gpu 0 --result0 ./result0/

There are other options you can choose.
Please refer to the code.

To use our pretrained model, please download here

Contact

If you have any question about the code or paper, please feel free to contact kkbbbj@gmail.com