Skip to content

camalab-ai/LVD

Repository files navigation

A Lightweight RAW Video Denoiser with Coarse Alignment and Imbalanced Noise Pre-training

Environment

The code runs on Python 3.8. You can create a virtualenv by running

virtualenv -p /usr/bin/python3.8 venv
source venv/bin/activate
pip install -r requirements.txt

Training

  1. Login to wandb from your console
  2. To train LVD model:
python train_model.py --init_model 1 --ofm --dr 6 --pe 10 --pf 16 --pc --gain \
	--trainset_dir <path_to_ReCRVD_dataset> \
	--valset_dir <path_to_ReCRVD_dataset> \
	--log_dir <path_to_dir_for_chkecpoints>

Quantitative evaluation:

python eval_models.py
    --data_dir <path_to_ReCRVD_dataset> \
    --model <path_to_model>

Data

ReCRVD dataset can be downloaded following instructions from the authors: https://github.com/cao-cong/RViDeformer

About

The official repo of the paper "A Lightweight RAW Video Denoiser with Coarse Alignment and Imbalanced Noise Pre-training"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages