• Corpus ID: 236447849

Adaptive Denoising via GainTuning

@inproceedings{Mohan2021AdaptiveDV,
  title={Adaptive Denoising via GainTuning},
  author={Sreyas Mohan and Joshua L Vincent and Ramon Manzorro and Peter A. Crozier and Eero P. Simoncelli and Carlos Fernandez-Granda},
  booktitle={Neural Information Processing Systems},
  year={2021}
}
Deep convolutional neural networks (CNNs) for image denoising are usually trained on large datasets. These models achieve the current state of the art, but they have difficulties generalizing when applied to data that deviate from the training distribution. Recent work has shown that it is possible to train denoisers on a single noisy image. These models adapt to the features of the test image, but their performance is limited by the small amount of information used to train them. Here we… 
[PDF] Semantic Reader
6 Citations
Background Citations
3
Methods Citations
3

6 Citations

Deep Denoising for Scientific Discovery: A Case Study in Electron Microscopy

This work proposes a simulation-based denoising (SBD) framework, in which CNNs are trained on simulated images, and tests the framework on transmission electron microscopy data, showing that it outperforms existing techniques on a simulated benchmark dataset, and on real data.

Patch-Craft Self-Supervised Training for Correlated Image Denoising

This work proposes a novel self-supervised training technique suitable for the removal of unknown correlated noise and evaluates the proposed framework through extensive experiments with synthetic and real image noise.

Test-Time Training Can Close the Natural Distribution Shift Performance Gap in Deep Learning Based Compressed Sensing

This work proposes a domain adaptation method for deep learning based compressive sensing that relies on self-supervision during training paired with test-time training at inference that essentially closes the distribution shift performance gap for state-of-the-art architectures for accelerated MRI.

Multi-rate adaptive transform coding for video compression

This study proposes learned transforms and entropy coding that may either serve as (non)linear drop-in replacements, or enhancements for linear transforms in existing codecs, and augmented the DCT with learned quantization matrices and adaptive entropy coding to compress intra-frame AV1 block prediction residuals.

Evaluating Unsupervised Denoising Requires Unsupervised Metrics

The proposed metrics enable unsupervised evaluation of denoising methods based exclusively on noisy data and are asymptotically consistent estimators of the supervised MSE and PSNR.

Statistical whitening of neural populations with gain-modulating interneurons

An online statistical whitening algorithm is derived that regulates the joint second-order statistics of a multi-dimensional input by adjusting the marginal variances of an overcomplete set of interneuron projections and can be generalized to achieve a form of local whitening in convolutional populations, such as those found throughout the visual or auditory system.

References

SHOWING 1-10 OF 86 REFERENCES

Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

This paper investigates the construction of feed-forward denoising convolutional neural networks (DnCNNs) to embrace the progress in very deep architecture, learning algorithm, and regularization method into image Denoising, and uses residual learning and batch normalization to speed up the training process as well as boost theDenoising performance.

Robust and interpretable blind image denoising via bias-free convolutional neural networks

It is shown that deep convolutional networks systematically overfit the noise levels for which they are trained: when deployed at noise levels outside the training range, performance degrades dramatically; in contrast, a bias-free architecture is shown, obtained by removing the constant terms in every layer of the network, including those used for batch normalization.

FFDNet: Toward a Fast and Flexible Solution for CNN-Based Image Denoising

The proposed FFDNet works on downsampled sub-images, achieving a good trade-off between inference speed and denoising performance, and enjoys several desirable properties, including the ability to handle a wide range of noise levels effectively with a single network.

Unsupervised Deep Video Denoising

An Unsupervised Deep Video Denoiser (UDVD1), a CNN architecture designed to be trained exclusively with noisy data, is proposed and the performance of UDVD is comparable to the supervised state-of-the-art, even when trained only on a single short noisy video.

Self2Self With Dropout: Learning Self-Supervised Denoising From Single Image

A self-supervised learning method which only uses the input noisy image itself for training, which significantly outperforms existing single-image learning or non-learning methods, and is competitive to the denoising networks trained on external datasets.

Probabilistic Noise2Void: Unsupervised Content-Aware Denoising

Probabilistic Noise2Void (PN2V) is presented, a method to train CNNs to predict per-pixel intensity distributions and a complete probabilistic model for the noisy observations and true signal in every pixel is obtained.

LIDIA: Lightweight Learned Image Denoising with Instance Adaptation

This work proposes a novel lightweight learnable architecture for image denoising, using a combination of supervised and unsupervised training of it, the first aiming for a universal denoiser and the second for an instance adaptation.

High-Quality Self-Supervised Deep Image Denoising

This work builds on a recent technique that removes the need for reference data by employing networks with a "blind spot" in the receptive field, and significantly improves two key aspects: image quality and training efficiency.

Deep Denoising for Scientific Discovery: A Case Study in Electron Microscopy

This work proposes a simulation-based denoising (SBD) framework, in which CNNs are trained on simulated images, and tests the framework on transmission electron microscopy data, showing that it outperforms existing techniques on a simulated benchmark dataset, and on real data.

How transferable are features in deep neural networks?

This paper quantifies the generality versus specificity of neurons in each layer of a deep convolutional neural network and reports a few surprising results, including that initializing a network with transferred features from almost any number of layers can produce a boost to generalization that lingers even after fine-tuning to the target dataset.
...