/MSLTNet

WACV 2024 (Official implementation of "4K-Resolution Photo Exposure Correction at 125 FPS with ~ 8K Parameters")

Primary LanguageJupyter Notebook

MSLTNet

WACV 2024 (Official implementation of "4K-Resolution Photo Exposure Correction at 125 FPS with ~ 8K Parameters") [arxiv] [Paper]

A quick glimpse of comparison on the ME dataset:

Dependencies and Installation

# create new anaconda env
conda create -n mslt python=3.7 -y
source activate mslt

# install python dependencies
pip3 install -r requirements.txt

Prepare Dataset

Download ME Dataset from [Baidu Disk] or [google drive] , unzip the file and put it in data/

Quick Inference

# Inference On ME Dataset
python SingleTest.py

Model Performance Testing

python test_model.py

Train MSLT

  python Train.py

Process Videos

  1. Copy the videos that you want to process using MSLT Network into sample_video folder.
  2. Activate the MSLT virtual environment.
  3. Run python video_process.py in the root directory.
  4. Output video will be inside sample_video/$video_name$/output_$video_name$ folder

Results Comparison

merged_Sony.SIII_1080p.mp4