Diabetic Retinopathy Detection project
Beijing University of Technology – 北京工业大学
[5 months engineering program]
The DRD project – Diabetic Retinopathy Detection project– was undertaken by Sajida Zouarhi under the guidance of Professor Yang within the Biotechnology Laboratory of BJUT – Beijing University of Technology.
The DRD is a research project in image processing. It aims to propose a segmentation solution of fundus images with minimum noise and minimum loss of information.
The analysis of the segmented image enables us to find signs of abnormalities on the blood vessel tree which makes early detection (and treatment) of the disease possible.
3D training for maintenance
Thales Airborne systems & Avionics
[6 months project]
How to train a maintenance operator when equipment is not available ?
- Our solution: a simulation tool (Unity 3D) with 3D object modelling and hand motion capture (Leap Motion).
- Fictitious Scenario: you have 2 minutes to defuse a bomb, you can use virtual tools that will appear in your virtual hand. If you touch the wrong wire the bomb explodes.
- Goal: the trainee will learn the sequence of actions to perform in the right order, recognize the tools that he has to use, know the precision that is needed for a specific action etc.
- Further improvement possible: haptic feedback.
Solution proposal for predictive maintenance on military vehicles
[4 months internship]
How to predict the RUL (remaining useful life) of an equipment in order to schedule in advance maintenance operations?
- State of art
- Functional architecture proposal (recommendation of sensors to deploy on vehicle, architecture to collect HUMS* measurments, algorithms to detect failure modes and estimate the RUL of the vehicle for smart maintenance scheduling)
- A model for reliability conversion for OPEX operations (using degradation law for mecanical and electronic equipment).
*HUMS: health and usage monitoring systems.
HomeNetwork – Home care system
ECE Paris (Engineering School)
[2 months project]
► HomeNetwork is home care system for elderly people. It’s a telemonitoring system that can detect the presence of a person in its home with a sensor network and trigger an alert in case of problem detection (fall).
The tool first “observe” the behaviour of the person during the learning phase (which is consider as normal behaviour), then the tool is automatically configured its alert threshold (unusual pattern recogntion) according to the person lifestyle.
► Tech: OS Contiki – COOJA – TELOS B (ZIGBEE 802.15.4) – Labview.