About me¶
I'm a second year Computer Vision PhD student at IMAGINE (ENPC) and GeoViC (Ecole Polytechnique) labs. My research is focused on conditional generative models. I've worked with semantically conditioned GANs and focused on subject transfer in images. Currently living in France, i am French, Spanish and Colombian. I'm bilingual in both French and Spanish while having C1 level in English
Supervisors¶
- David PICARD (LIGM, École des Ponts, Univ Gustave Eiffel, CNRS, France)
- Vincent LEPETIT (LIGM, École des Ponts, Univ Gustave Eiffel, CNRS, France)
- Vicky KALOGEITON (LIX, École Polytechnique, France)
Publications¶
Nicolas Dufour, Vicky Kalogeiton, David Picard
Paper | Webpage | Code
Published at: ECCV 2022
In this work, we introduce SCAM (Semantic Cross Attention Modulation), a system that encodes rich and diverse information in each semantic region of the image (including foreground and background), thus achieving precise generation with emphasis on fine details.
We use this new architecture to perform subject transfer and set a new state of the art for this task.
Open Source Contributions¶
- TorchRL: Worked on the model based aspect of the library
- torchmetrics: Improved the FID metric to work online, avoiding storing features.
- huggingface/diffusers: Implemented the sampling for K-LMS sampler
Short CV¶
July 2022 - October 2022 | Internship at Meta London on Model Based Reinforcement Learning. Working in the Pytorch team on TorchRL |
October 2021 - now | PhD student on "Generation and comprehension of dynamic scenes." supervised by Vicky KALOGEITON, David PICARD and Vicky KALOGEITON at IMAGINE / ENPC & GeoViC / École Polytechnique |
April 2021 - September 2021 | Research internship on "Subject Transfer and Conditional Generative Adversarial Networks" supervised by David PICARD and Vicky KALOGEITON at IMAGINE / ENPC & GeoViC / École Polytechnique |
September 2020- September 2021 | Mathematics, Vision and Learning (MVA) master at ENS Paris Saclay |
January 2020-August 2020 | Machine Leaning Engineer Intern at Upskills Singapore. Worked on rogue trading detection using Graph Neural Networks |
September 2017 - September 2020 | Engineering degree at Telecom SudParis. Followed the MSA specialty (Mathematics and Statistical Modeling) |
More detailed cv here