Image Enhancement

This is the source code for image enhancement tasks. Currently, it supports rain removal in daytime conditions.

Installation

Requirements

• Anaconda3
• Python 3.9
• CUDA 11.1
• PyTorch 1.10
• Code has been tested on Ubuntu 20.04 / Windows 10

Setup Environment

• Download the whole source code.
• Goto setup folder cd image_enhancement/setup
• Create the Anaconda environment: conda env create -f mlkit.yml

Training

• Download the training data from: [download](https://o365skku-my.sharepoint. com/:u:/g/personal/phlong_o365_skku_edu/ETZ4XCf9oxhEvfhrchrXXZwBecAZaP1YFBBzrGwQlwM5Kw?e=TtQfeL) (~7 GB).
• Extract the data to image_enhancement/data.
  • It should be located at: image_enhancement/data/rain
• Run the training scripts: python image_enhancement/exps/run/train.py

Inference

• If you have retrained the model, find the best weight from: image_enhancement/exps/checkpoints/mprnet/mprnet_rain/<version>/weights /best...ckpt
• Copy the best weight to image_enhancement/models_zoo. Rename it as: mprnet_rain_version_0.ckpt
• Run the inference scripts: python image_enhancement/exps/run/infer.py