Minimal implementation of Deep Joint Demosaicking and Denoising [Gharbi2016]
Project description
# Deep Joint Demosaicking and Denoising SiGGRAPH Asia 2016
Michaël Gharbi gharbi@mit.edu Gaurav Chaurasia Sylvain Paris Frédo Durand
A minimal pytorch implementation of “Deep Joint Demosaicking and Denoising” [Gharbi2016]
# Installation
From this repo:
`shell python setup.py install `
Using pip:
`shell pip install demosaicnet `
Then run the demo script with:
`shell python scripts/demosaicnet_demo.py output `
To train a dummy model on the demo dataset provided, run:
`shell python scripts/train.py --data demosaicnet/data/dummy_dataset --checkpoint_dir ckpt `
To build and update the whee:
`shell pip install wheel twine make distribution make upload_distribution `
# FAQ
How is noise handled? Where is the pretrained model? The noise-aware model is not implementation, see the earlier Caffe implementation for that <https://github.com/mgharbi/demosaicnet_caffe>
How do I train this? The script scripts/train.py is a good start to setup your training job, but I haven’t tested it yet, I recommend rolling your own.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for demosaicnet-0.0.14.macosx-12.4-arm64.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | f7a8126c0366b84acdfab1f8487a7aceeeada7f8926547a6053740546041b095 |
|
MD5 | ec6b728c8f59e50f06a0bc7577a21484 |
|
BLAKE2b-256 | 4c0da3f8ddbc2ddd0ff2ae18987c695c26b1658a45237812504d520cbad925d3 |
Hashes for demosaicnet-0.0.14-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | e50d502a1336736fc83fdaca97c6e709c3114419083cb1854def3355fb6894c9 |
|
MD5 | 9dc4b56881feb01725ce99e23f0559a4 |
|
BLAKE2b-256 | 1e87048de313cecc62268dbd7c5d9f35ea34887e84330bb4d8d03610c63eef5b |