Self2Self With Dropout: Learning Self-Supervised Denoising From Single Image
By Yuhui Quan, Mingqin Chen, Tongyao Pang, Hui Ji. Computer Vision and Pattern Recognition, 2020
456 citations recorded by the source corpus.
In last few years, supervised deep learning has emerged as one powerful tool for image denoising, which trains a denoising network over an external dataset of noisy/clean image pairs. However, the requirement on a high-quality training dataset limits the broad applicability of the denoising networks. Recently, there have been a few works that allow training a denoising network on the set of external noisy images only. Taking one step further, this paper proposes a self-supervised learning method which only uses the input noisy image itself for training. In the proposed method, the network is trained with dropout on the pairs of Bernoulli-sampled instances of the input image, and the result is estimated by averaging the predictions generated from multiple instances of the trained model with dropout. The experiments show that the proposed method not only significantly outperforms existing single-image learning or non-learning methods, but also is competitive to the denoising networks trained on external datasets.
Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition, Media Technology
View Paper: https://www.semanticscholar.org/paper/68a7b1a8dc78b4d9d7fca1b6da1dcb3d5d5285bc
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Biomedical Engineering
- Richard Baraniuk
- Huchuan Lu
- Xu Jia
- Guha Balakrishnan
- Marco F. Duarte
- Steven L. Brunton
- J. Nathan Kutz
- Justin K. Romberg
In Computer Vision and Pattern Recognition
- Huchuan Lu
- Xu Jia
- Jian-zhuo Liu
- Richard Baraniuk
- Ashok Narayanan Veeraraghavan
- Daniel LeJeune
- Songjiang Li
- Tao Huang
In Media Technology
- Jian-zhuo Liu
- Richard Baraniuk
- Ashok Narayanan Veeraraghavan
- Songjiang Li
- Huchuan Lu
- Xu Jia
- Tao Huang
- Guha Balakrishnan
Citations in the corpus
- Deep learning models for digital image processing: a review (2024, unknown): “Quan et al. (2020) introduced a self-supervised deep learning method named Self2Self for image denoising.”
- Neighbor2Neighbor: Self-Supervised Denoising from Single Noisy Images (2021, background): “For deep image denoisers trained with only a single noisy image, Ulyanov et al. [29] proposed deep image prior (DIP), where the image prior is captured from the neural network rather than specially designed; Self2Self [25] and Noisy-as-Clean [31] are some recent works.”
- WIRE: Wavelet Implicit Neural Representations (2023, result): “We also compared the denoising result with state-of-the-art self-supervised denoising technique, “self2self [32].”