Installation and train DerainNet

Installation

Run the script below to install the environment (Conda):

conda create -n derainnet python=3.7
conda activate derainnet
conda install -y numpy=1.19.5 tensorflow-gpu=1.15 matplotlib
pip install scikit-image==0.17.2

Datasets

To train the models, please download the synthetic datasets (Rain100L, Rain100H, Rain1400) and and place the unzipped folders into the project folder: https://drive.google.com/file/d/1PES0IFPQ24MxpP1_6f_TvSBw4U8KdK_E/view?usp=sharing

Getting Started

1) Testing

Run shell scripts to test the models:

bash test.sh

You can modify the arguments in the test.sh or testing.py

2) Training

Run shell scripts to train the models:

bash train.sh      

The models are saved into ./model/ folder by default You can modify the arguments in the training.sh, training.py or network.py