Wael Saideni

Mr. Wael Saideni

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 March 2022 in Applied Sciences
Verified
Journal Article
Open access
PDF file
WSWael Saideni
Wael Saideni
DHDavid Helbert
David Helbert
FCFabien Courreges
Fabien Courreges
JCJean-Pierre Cances
Jean-Pierre Cances
The use of compressive sensing in several applications has allowed to capture impressive results, especially in various applications such as image and video processing and it has become a promising direction of scientific research. It provides extensive application value in optimizing video surveillance networks. In this paper, we introduce recent state-of-the-art video compressive sensing methods based on neural networks and categorize them into different categories. We compare these approaches by analyzing the networks architectures. Then, we present their pros and cons. The general conclusion of the paper identify open research challenges and point out future research directions. The goal of this paper is to overview the current approaches in image and video compressive sensing and demonstrate their powerful impact in computer vision when using well designed compressive sensing algorithms.


1 January 2021 in HAL
Verified
Preprint
Open access
PDF file
WSWael Saideni
Wael Saideni
FCFabien Courreges
Fabien Courreges
DHDavid Helbert
David Helbert
JCJean Pierre Cances
Jean Pierre Cances
The use of recurrent neural networks in several applications has allowed to capture impressive results, especially in various applications such as video prediction and it has become a promising direction of scientific research. In this paper, we introduce a novel algorithm for video prediction called "Robust Spatiotemporal Convolutional Long Short-Term Memory (Robust-ST-ConvLSTM) Algorithm" that outperforms the state-of-the-art approaches. Robust-ST-ConvLSTM is a memory flow algorithm based on higher order ConvLSTM. This memory flow algorithm is holding the spatiotemporal information to optimize and control the prediction abilities of the ConvLSTM cell. Our approach is developed in the specific context of predicting future frames based on historical observations, and we experimentally validate the ability of the proposed algorithm on two spatiotemporal datasets, including a moving variant of MNIST dataset of handwritten digits, and KTH which is a human motion dataset.

7 April 2026
Verified
Thesis
Closed access
WSWael Saideni
Wael Saideni
Optimisation d'un système de video surveillance dans les villes intelligentes en utilisant le Deep Learning et la video compressive sensing: La technique de compressive sensing joue un rôle important dans le traitement des données vu que l’acquisition et la compression se font simultanément grâce à un processus de prise de mesures. Cette technique optimise les capacités de stockage des systèmes ainsi que la vitesse et le coût d’acquisition. Récemment, cette technique est devenue de plus en plus utilisée grâce à l’optimisation des algorithmes de reconstruction en utilisant les architectures du Deep Learning. L’objectif principal de cette thèse est de tirer profit des architectures de Deep Learning pour optimiser la technique de compressive sensing en l’appliquant sur des signaux vidéo et par la suite optimiser l’acquisition, la transmission et la reconstruction des vidéos dans les systèmes numériques modernes. Ainsi, la stratégie adoptée au cours de ces travaux de recherche consiste à commencer par établir une étude comparative sur les approches de vidéo compressive sensing (VCS) basées sur le Deep Learning en évaluant la qualité et la vitesse de reconstruction ainsi que les différentes architectures. Puis, deux environnements de VCS ont été conçus : le premier se base sur la prédiction des frames vidéo en implémentant une approche basée sur un nouveau réseau récurrent et le deuxième exploite les dernières performances réalisées avec les Transformers et le mécanisme d’attention. Alors, la démarche adoptée dans ces deux approches repose sur une analyse de l’état de l’art suivie d’une explication de chaque architecture et une validation expérimentale. Les différentes contraintes rencontrées au cours de ces travaux sont discutées et des solutions appropriées sont proposées.

13 December 2022 in HAL
Verified
Preprint
Open access
PDF file
WSWael Saideni
Wael Saideni
FCFabien Courreges
Fabien Courreges
DHDavid Helbert
David Helbert
JCJean Pierre Cances
Jean Pierre Cances
In this work, we study a complete framework of Video Compressive Sensing (VCS), from capturing a sequence of video frames in one single compressed measurement to reconstructing the original frames. To our best knowledge, we present the first end-to-end sampling and recovery network built upon video Transformers, widely explored in vision related tasks, to capture long-range spatio-temporal relations. Our proposedVideo Transformer for Snapshot Compressive Imaging recovery (ViT-SCI) is based on Spatio-temporal Conventional Multi-HeadAttention (ST-ConvMHA) which is an extended version of the fully-connected attention adapted for vision problems. Our comprehensive qualitative and quantitative experiments on several datasets demonstrate that ViT-SCI outperforms previous state-of-the-art methods with much faster reconstruction capacities, which pave the way for applying our algorithm in real-time applications. Indeed, ViT-SCI achieves good quality results inonly 8ms to reconstruct video clips of 8 frames. It is able to perform real-time reconstruction of up to 125 measurements per second. Finally, an important ablation study on the Transformer network is provided to inspire future research works aiming to test the abilities of Transformers in vision tasks.