Burst Denoising with Kernel Prediction Networks

Ben Mildenhall | Jonathan T. Barron | Jiawen Chen | Dillon Sharlet | Ren Ng | Robert Carroll

This is not an official Google product. This repository contains code for training models from the paper Burst Denoising with Kernel Prediction Networks.

Dependencies

This code uses the following external packages:

• TensorFlow
• NumPy
• SciPy
• Matplotlib

Dataset

Synthetic training data is generated using the OpenImages dataset, which can be manually downloaded following these instructions.

Training

Run the following command to train the kernel prediction network (KPN) burst denoising model:

python kpn_train.py --dataset_dir $OPEN_IMAGES_DATASET_DIR --data_dir $REAL_BURST_DATA_DIR