Othman Sbai

Computer vision researcher

I am a PhD student at Ecole des Ponts and AI resident @ Facebook AI Research (FAIR), where I work on computer vision and machine learning.

I am advised by Mathieu Aubry from École des Ponts and Camille Couprie from FAIR.


I'm interested in building computer vision algorithms that empower impactful products.

Research

I'm interested in computer vision and machine learning.

During my PhD, I have worked on building algorithms designed for creative applications such as image generation and editing particularly for fashion, and visual metaphors creation.

I have also worked on learning image representations suited for image retrieval and few-shot classification.

Surprising image compositions

Othman Sbai, Mathieu Aubry, Camille Couprie

Accepted in CVPR 2021 Workshop on fashion, art and design and ICCC 2021

We build an algorithm that suggests visual metaphors based on the visual similarity of objects and their contexts and performs the copy paste seamlessly.

Impact of base dataset design on few-shot classification

We study how the training set labelling impacts the quality of obtained features.

Unsupervised image decomposition in vector layers

Unsupervised decomposition of images as superimposed alpha-channel vector masks for image editing.

Design inspiration from generative networks

Can an algorithm create original and compelling fashion designs to serve as an inspirational assistant ? To help answer this question, we design and investigate different image generation models and ways to improve novelty.

Education

2018 - 2021

Ph.D. in computer vision and machine learning

Facebook AI Research & Ecole des Ponts


2016 - 2017

MSc. MVA - Applied Maths to Vision and Language

Ecole des Ponts & ENS Cachan, France


2013 - 2017

Master of engineering in computer science

Ecole des Ponts et Chaussées, France

2011 - 2013

Advanced Mathematics, physics and engineering

Moulay Youssef CPGE, Rabat, Morocco


Previous experiences

Sport Shooting

Sep 2015-Jul 2016 - Renault Nissan Silicon Valley Research Center

For racing fans, having an aerial video of their performance is exciting.

We built an iOS app for following and filming a car on a race track using a DJI Inspire 1 drone.

From real time trajectory planning, to drone flight control and visual tracking of the car, it was an exciting and challenging project that I had the opportunity to work on for 10 months.

KriDari - Vacation home rental platform

One of the big worries when planning holidays is finding a convenient housing in a short time. This is even more challenging in the peak of tourism season.

For that, we built KriDari.ma as the first Moroccan platform for vacation home rental.

KriDari Fb page - Demo video

Skills

Languages

  • English (Fluent)

  • French (Bilingual)

  • Arabic (Native)

  • German (Beginner)


Coding

  • Python, C++, Javascript, Objective-C

  • Machine learning: PyTorch, Tensorflow, Matlab, R

  • HTML, CSS, Node.js, Express, Ruby-on-Rails, php