Deep Bilateral Learning for Real-Time Image Enhancements

Unofficial PyTorch implementation of 'Deep Bilateral Learning for Real-Time Image Enhancement', SIGGRAPH 2017 https://groups.csail.mit.edu/graphics/hdrnet/

Python 3.6

Dependencies

To install the Python dependencies, run:

pip install -r requirements.txt

Datasets

HDR+ Burst Photography Dataset - https://hdrplusdata.org/dataset.html

Getting the data

To get started, using the subset of bursts (153 bursts, 37 GiB).

gsutil -m cp -r gs://hdrplusdata/20171106_subset .

Usage

To train a model, run the following command:

python train.py 
--raw_path="/content/drive/My Drive/HDR+ Dataset/20171106_subset/results_20171023/*/merged.dng"
--hdr_path="/content/drive/My Drive/HDR+ Dataset/20171106_subset/results_20171023/*/final.jpg"

To test image run:

python inference.py --pretrain_dir="weights//ckpt" --input_path=<raw image path> --output_path=<saved image path>

Known issues