File size: 1,330 Bytes
c4d15b6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
---
license: bsd-3-clause
library_name: pytorch
tags:
  - image-denoising
  - self-supervised-learning
  - deepinv
  - arxiv:2603.25869
---

# Learning to Recorrupt (L2R)

**Noise distribution agnostic self-supervised image denoising.**

Brayan Monroy 路 Jorge Bacca 路 Julian Tachella

[Paper](https://arxiv.org/abs/2603.25869) 路 [Code](https://github.com/bemc22/learning2recorrupt) 路 [Weights](https://huggingface.co/bemc22/learning2recorrupt/tree/main)

L2R jointly learns a denoiser and a monotonic recorruptor using noisy images alone.

### Weights

Eight RGB DRUNet checkpoints cover log-gamma, Laplace, correlated Gaussian,
Poisson鈥揋aussian, and Poisson noise. [manifest.json](manifest.json) contains their
settings and SHA-256 hashes. [mfcn_init.pth](mfcn_init.pth) initializes the monotonic
network as a Gaussian identity map.

### Use

From the code repository, with dependencies installed and test caches available:

```bash
hf download bemc22/learning2recorrupt --local-dir pretrained
python test.py --checkpoint-dir pretrained
```

### Citation

```bibtex
@article{monroy2026learning,
  title={Learning to Recorrupt: Noise Distribution Agnostic Self-Supervised Image Denoising},
  author={Monroy, Brayan and Bacca, Jorge and Tachella, Juli{\'a}n},
  journal={arXiv preprint arXiv:2603.25869},
  year={2026}
}
```