Stéphane Sochacki

A new collaboration with A2Drone, this time for the Grand Port Maritime de La Rochelle, for the detection and monitoring of defects on the port's facilities, particularly the quays (concrete and masonry) and concrete pillars.

I contributed my expertise in image processing and analysis and deep learning (C++/OpenCV/YOLOV4) by leveraging the generic image analysis platform I previously developed.

I also participated in the implementation of 3D photogrammetry solutions for better visualization of defects in their environment.

This collaboration with A2Drone for the company Enedis addressed the issue of defect detection on high-voltage power lines using drone-acquired videos.

During this assignment, I applied my expertise as a software engineer by designing and developing (C++/OpenCV) a generic image analysis platform.

I also contributed my expertise in image processing and analysis by developing methods suited to our problem (background suppression, color segmentation, etc.).

Finally, I developed the artificial intelligence layer (convolutional neural networks) for the detection and tracking of infrastructure components, as well as for their diagnosis.

As Technical Director, I was responsible for recruitment and management, information system design, technological choices, research (artificial intelligence, NLP, data mining, content detection in images), and development (C#, MSSQL, OpenCV).

To meet the diverse needs of projects and clients (BFMTV, Next Interactive Media, Viacom/MTV, Ouest France, National Union of Professional Footballers), I designed a generic curation system focused on collecting (press articles, social networks), storing, and analyzing heterogeneous data to serve the results of these analyses via a REST API for presentation on thin clients (website or mobile app).