Burst image denoising using a RNN based CNN architechture.

This is the re-implement of the paper End-to-End Denoising of Dark Burst Images Using Recurrent Fully Convolutional Networks accessing at https://arxiv.org/abs/1904.07483.

TODO：upload the code of training on the Vimeo-90K dataset

All codes are implemented by PyTorch 1.0.0 and Numpy.

In my simulations, the dataset selected is CID Dataset (Learning to See in the Dark).

Before to train a model, the train_list.txt and val_list.txt should be generated by the python code generate_list.py, and this code has included in the train code 'rnn_fcn_train.py'.

If you like the codes or they are helpful for you, give a Start or Fork to support me.