https://github.com/sadevans/b2u_sber_implemetation
Модификация Blind2Unblind для конкретных данных и задач
https://github.com/sadevans/b2u_sber_implemetation
blind2unblind computer-vision opencv pytorch
Last synced: 3 months ago
JSON representation
Модификация Blind2Unblind для конкретных данных и задач
- Host: GitHub
- URL: https://github.com/sadevans/b2u_sber_implemetation
- Owner: sadevans
- Created: 2023-07-27T08:51:47.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2023-07-27T14:43:23.000Z (almost 2 years ago)
- Last Synced: 2025-01-19T00:46:35.920Z (5 months ago)
- Topics: blind2unblind, computer-vision, opencv, pytorch
- Language: Jupyter Notebook
- Homepage:
- Size: 15.2 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# That is implementation of Blind2Unblind: Self-Supervised Image Denoising with Visible Blind Spots for SBER Robotics Lab
article: [Blind2Unblind](https://arxiv.org/abs/2203.06967)
## Citing Blind2Unblind
```
@InProceedings{Wang_2022_CVPR,
author = {Wang, Zejin and Liu, Jiazheng and Li, Guoqing and Han, Hua},
title = {Blind2Unblind: Self-Supervised Image Denoising With Visible Blind Spots},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2022},
pages = {2027-2036}
}
```
The original code is placed here: [github](https://github.com/zejinwang/Blind2Unblind)
## Installation
The model is built in Python3.8.5, PyTorch 1.7.1 in Ubuntu 22.04 environment.
## Data Preparation
### 1. Prepare Training Dataset
Please put your training dataset under the path: **./b2u_sber_implemetation/data/train**.
### 2. Prepare Validation Dataset
Please put your validation dataset under the path: **./b2u_sber_implemetation/data/test**.
## Pretrained Models
You can find pre-trained models here: **./b2u_sber_implemetation/pretrained_models**
Models were trained on datasets G-209, Crystal_focus_0_dose_180, G-146
```yaml
# # For more noisy datasets processing use model firstly trained on G-209
./pretrained_models/b2u_first.pth
# Than use model secondly trained on G-209 denoised by first model
./pretrained_models/b2u_second.pth
# # For less noisy images use model trained on Crystal_focus_0_dose_180
./pretrained_models/b2u_crystal_first.pth
```
## Train
* For training your own model please use [SBER_train](https://github.com/sadevans/b2u_sber_implemetation/blob/f2865e86ba95634329dfbdb229182295d3da0425/SBER_train.ipynb#L10)
## Test
Please put your test data in the folder: **./b2u_sber_implemetation/test**
* To test model on images maximum size **768x1024** use [SBER_test_small_images](https://github.com/sadevans/b2u_sber_implemetation/blob/main/SBER_test_small_images.ipynb)
* To test model on large resolution images use [SBER_test_large_images](https://github.com/sadevans/b2u_sber_implemetation/blob/main/SBER_test_large_images.ipynb)
In this jupyter notebook you can set:
- your image proportions,
- crop propotions
- margin value for cropping and concating without visible joints