Brief Biography

Background and Position

  • 2016 - present PhD candidate in the Université Côte d'Azur at the laboratory CNRS-I3S. Research engineer at Airbus (CIFRE scholarship).
    Title: Dense classification and segmentation using RGB-D sensors for the accurate localisation of a manufacturing collaborative robot.
  • 2015 (6 months) Computer vision Research intern at Dassault Systèmes, Corporate Research.
    Title: Real-time multimodal tracking on mobile devices for augmented reality applications.
  • 2014 (6 months) Hardware electrical engineering intern at SYSNAV.
    Title: Conception and development of a 3D high-precision magnetic field generator controlled by digital profiles.
  • 2014 - 2015 Master of Science (Msc) in Digital Signal and Image Processing from INSA de Lyon, France.
  • 2012 - 2015 Engineering degree and Master of Science (MSc) majoring in Electrical and Computer Engineering from INSA de Lyon, France.
  • 2009 - 2012 Bachelor of Science (BSc) majoring in Aerospace Engineering from ESTACA, France.

Research Areas

My research interests lie at the intersection of computer vision, machine learning and robotics with a focus on scene understanding for robots evolving in environments modeled by pre-existing 3D models. Topics of interest include deep neural networks, semantic segmentation, semantic instance segmentation, visual odometry, real-time direct tracking, visual SLAM.

I'm currently working on COMANOID project (2015-2019) - for Multi-Contact Collaborative Humanoids in Aircraft Manufacturing - which is a 4-year European research project as part of the Horizon H2020 program.



  1. "Classification dense d'images de profondeur entraînée sans données réelles", Howard Mahé, Denis Marraud, Andrew I. Comport, 
    XXVIème colloque GRETSI (GRETSI 2017), Sep 2017, Juan-les-Pins, France. 2017



  • IP Days 2016 @ Airbus Defense & Space, Toulouse : Deep Learning for Semantic Scene Understanding (21/11/2016) [slides, poster]
  • IP Hours 2016 @ Airbus Group Innovations, Suresnes : Conditional Random Fields in Computer Vision (15/04/2016) [slides]


  • "Learning Segmentation of Depth maps from 3D CAD Models" [poster], Howard Mahé, Denis Marraud, Andrew I. Comport,
    International Computer Vision Summer School (ICVSS), Jul 2017, Sicily, Italy. 2017