Raphael Achddou

2 Bd Blaise Pascal

93160 Noisy-le-Grand, France

I am an Assistant Professor (Maitre de Conférences) teaching at ESIEE Paris and affiliated with the LIGM laboratory, in the A3SI team. My main research interests are computer vision, image processing, and the statistical modelling of natural images. In particular, I work on solving imaging inverse problems with deep learning.

Before joining ESIEE, I was a postdoctoral researcher at EPFL working under the supervision of Sabine Süsstrunk in the IVRL laboratory. I obtained my PhD in Signal and Data Processing from the Institut Polytechnique de Paris, with Pr.Yann Gousseau and Pr.Saïd Ladjal. During my PhD, I was affiliated with the IMAGES team of Télécom Paris LTCI lab.

news

Mar 15, 2026	Our paper 2-Shots in the Dark: Low-Light Denoising with Minimal Data Acquisition was accepted at CVPR 2026!
Oct 07, 2025	I will give a talk at the Imaging In Paris seminar! slides
Sep 15, 2025	I will join ESIEE Paris as a Maitre de Conférences in October!
Jul 23, 2024	I had the pleasure to co-organize ICCP 2024, the premier annual conference on Computational Imaging!
May 01, 2023	I recently joined the IVRL lab of professor Sabine Süsstrunk as a Post-Doctoral researcher.

selected publications

2-Shots in the Dark: Low-Light Denoising with Minimal Data Acquisition

Liying Lu, Raphael Achddou, and Sabine Susstrunk

2025

Abs arXiv

Raw images taken in low-light conditions are very noisy due to low photon count and sensor noise. Learning-based denoisers have the potential to reconstruct high-quality images. For training, however, these denoisers require large paired datasets of clean and noisy images, which are difficult to collect. Noise synthesis is an alternative to large-scale data acquisition: given a clean image, we can synthesize a realistic noisy counterpart. In this work, we propose a general and practical noise synthesis method that requires only one single noisy image and one single dark frame per ISO setting. We represent signal-dependent noise with a Poisson distribution and introduce a Fourier-domain spectral sampling algorithm to accurately model signal-independent noise. The latter generates diverse noise realizations that maintain the spatial and statistical properties of real sensor noise. As opposed to competing approaches, our method neither relies on simplified parametric models nor on large sets of clean-noisy image pairs. Our synthesis method is not only accurate and practical, it also leads to state-of-the-art performances on multiple low-light denoising benchmarks
VibrantLeaves: A principled parametric image generator for training deep restoration models

Raphael Achddou, Yann Gousseau, Saïd Ladjal, and 1 more author

2025

Abs arXiv

Even though Deep Neural Networks are extremely powerful for image restoration tasks, they have several limitations. They are poorly understood and suffer from strong biases inherited from the training sets. One way to address these shortcomings is to have a better control over the training sets, in particular by using synthetic sets. In this paper, we propose a synthetic image generator relying on a few simple principles. In particular, we focus on geometric modeling, textures, and a simple modeling of image acquisition. These properties, integrated in a classical Dead Leaves model, enable the creation of efficient training sets. Standard image denoising and super-resolution networks can be trained on such datasets, reaching performance almost on par with training on natural image datasets. As a first step towards explainability, we provide a careful analysis of the considered principles, identifying which image properties are necessary to obtain good performances. Besides, such training also yields better robustness to various geometric and radiometric perturbations of the test sets.