Today
Algorithm & AI engineer
LYNRED · Permanent role
Computer vision, deep learning and multimodal perception.
Grenoble, France
Grenoble, France
Yoann Dupas — algorithm & AI engineer.
I’m an algorithm & AI engineer at LYNRED, based in Grenoble, France. I bring together computer vision, deep learning and multimodal perception to better understand images.
My background connects doctoral research with applied engineering. Here you’ll find my experience, skills and publications.
My experience.
Previous experience
2022 — 2026
Doctoral research · CIFRE
Multimodal early fusion with attention mechanisms for pedestrian and vehicle detection.
2021 — 2022
Research & development engineer
WiseBIM · Chambéry
Research and development of document recognition solutions.
Mar — Aug 2021
Final-year engineering internship · Deep learning
WiseBIM · Chambéry
CNN development with Keras and TensorFlow; creation and annotation of training datasets.
2020
Research internship
I3S · Université Côte d’Azur
Research and development of deep neural networks, including LSTM and graph attention networks.
What I do.
Deep learning
PyTorch · TensorFlow · Keras · CNN · Transformers · LSTM · GAT
Computer vision
RGB / IR / LiDAR · YOLOv8 · RT-DETR · Attention · Segmentation
Engineering
Python · C++ · Docker · Git · CI/CD · MLOps · Quantization
My education.
2022 — 2026
PhD in Computer Science
Université Grenoble Alpes · MSTII
2018 — 2021
Engineering degree in Computer Science
Polytech Tours
2016 — 2018
Preparatory engineering cycle
Polytech Tours
A look at my research.
Two papers and a thesis on image fusion and multimodal perception.
MEFA: Multimodal Image Early Fusion with Attention Module for Pedestrian and Vehicle Detection
MEFA uses attention to fuse visible, infrared and LiDAR images before detection. It is studied with YOLOv8 and RT-DETR, including rainy and foggy conditions.
First author
MEFA-MS: attention-based U-Net for pedestrian and vehicle detection
MEFA-MS revisits early fusion with a U-Net architecture and spatial and channel attention. It aims to preserve detection quality while reducing inference cost.
First author
Advanced early fusion approaches for multimodal images in pedestrian and vehicle detection
My doctoral research investigates early multimodal image fusion for pedestrian and vehicle detection, within a CIFRE collaboration between Orange and LIG.
Doctoral research · Orange / LIG
A little more about me.
Outside research, you’ll find me hiking, climbing, cooking, playing video games or exploring the world of FPV drones. Different ways of experimenting, learning and seeing things from a new angle.
- Hiking
- Climbing
- FPV drones
- Cooking
- Video games
LET’S CONNECT
Let’s connect.
For scientific and technical discussions about computer vision and multimodal perception.
Send me an email