Search Results for author: Pablo Arias

Found 13 papers, 8 papers with code

SING: A Plug-and-Play DNN Learning Technique

1 code implementation25 May 2023 Adrien Courtois, Damien Scieur, Jean-Michel Morel, Pablo Arias, Thomas Eboli

We propose SING (StabIlized and Normalized Gradient), a plug-and-play technique that improves the stability and generalization of the Adam(W) optimizer.

Depth Estimation Image Classification

Can neural networks extrapolate? Discussion of a theorem by Pedro Domingos

no code implementations7 Nov 2022 Adrien Courtois, Jean-Michel Morel, Pablo Arias

In particular, what are the interpolation capabilities of trained neural networks?

Self-Supervised Super-Resolution for Multi-Exposure Push-Frame Satellites

no code implementations CVPR 2022 Ngoc Long Nguyen, Jérémy Anger, Axel Davy, Pablo Arias, Gabriele Facciolo

Modern Earth observation satellites capture multi-exposure bursts of push-frame images that can be super-resolved via computational means.

Earth Observation Super-Resolution

Self-supervision versus synthetic datasets: which is the lesser evil in the context of video denoising?

no code implementations25 Apr 2022 Valéry Dewil, Aranud Barral, Gabriele Facciolo, Pablo Arias

In this paper, we propose a study aiming to determine which is the best approach to train denoising networks for real raw videos: supervision on synthetic realistic data or self-supervision on real data.

Denoising Video Denoising

Investigating Neural Architectures by Synthetic Dataset Design

1 code implementation23 Apr 2022 Adrien Courtois, Jean-Michel Morel, Pablo Arias

It is therefore impossible to know what cues a given neural structure is taking advantage of in such data.

Depth Estimation

Proba-V-ref: Repurposing the Proba-V challenge for reference-aware super resolution

1 code implementation25 Jan 2021 Ngoc Long Nguyen, Jérémy Anger, Axel Davy, Pablo Arias, Gabriele Facciolo

We argue that in doing so, the challenge ranks the proposed methods not only by their MISR performance, but mainly by the heuristics used to guess which image in the series is the most similar to the high-resolution target.

Image Super-Resolution

Implementation of the VBM3D Video Denoising Method and Some Variants

1 code implementation6 Jan 2020 Thibaud Ehret, Pablo Arias

VBM3D is an extension to video of the well known image denoising algorithm BM3D, which takes advantage of the sparse representation of stacks of similar patches in a transform domain.

Image Denoising Optical Flow Estimation +1

Joint Demosaicking and Denoising by Fine-Tuning of Bursts of Raw Images

1 code implementation ICCV 2019 Thibaud Ehret, Axel Davy, Pablo Arias, Gabriele Facciolo

Due to the unavailability of ground truth data these networks cannot be currently trained using real RAW images.

Demosaicking Denoising

Model-blind Video Denoising Via Frame-to-frame Training

1 code implementation CVPR 2019 Thibaud Ehret, Axel Davy, Jean-Michel Morel, Gabriele Facciolo, Pablo Arias

Modeling the processing chain that has produced a video is a difficult reverse engineering task, even when the camera is available.

Denoising Video Denoising

Non-Local Video Denoising by CNN

2 code implementations30 Nov 2018 Axel Davy, Thibaud Ehret, Jean-Michel Morel, Pablo Arias, Gabriele Facciolo

To the best of our knowledge, this is the first successful application of a CNN to video denoising.

Image Denoising Video Denoising

On the Convergence of PatchMatch and Its Variants

no code implementations CVPR 2018 Thibaud Ehret, Pablo Arias

We also derive more specific bounds for two of these particular cases: the original PatchMatch and Coherency Sensitive Hashing.

Cannot find the paper you are looking for? You can Submit a new open access paper.