Open Computer Vision (OpenCV) est une bibliothèque graphique libre distribuée sous licence BSD. Elle est dédiée au traitement d’images en temps réel.
Context : vision system, that acquires images of faces under various modalities of imaging.
Missions and achievements :
- development of algorithms for the automatic diagnostic of skin : texture, visible pores and gloss.
- integration , deployment , on multiple OS, with cross-testing, on cloud (GCP) and on device ( taking into account cpu/memory/storage constraints).
- audit of algorithms, with respect to standard of development, and repeatability of training of models
- evolutive maintenance of solutions with bakers of the project, with respect to the use-case of the solution.
Context : web application for a social network of memes
Missions and achievements : Development of a dockerized web application that contains an algorithm for object-tracking on videos.
Context :X-ray imaging applied to threat detection
Missions and achievements :
- prototyping, integration and maintenance of algorithms :
Subjects : image reconstruction, image registering and unsupervised classification of images for anomaly detection, image segmentation for high energy discrimination
- maintenance of legacy projects/code : conversion and porting of matlab code to python
Technical environment : developpement : matlab , python,opencv,scikit-image,scikit-learn. compilation tools : nuitka, pyinstaller
Context : imaging software on android device for processing of images of documents.
Missions and achievements :
-prototyping and integration of software : - blur estimation on image acquired by a scanner, detection of holographic pattern, auto-exposure, autofocus ,restoration and binarization of images of documents acquired by a scanner.
Context : application (algorithmic pipeline) of 3D reconstruction , for satelite imaging.
Missions and achievements :
- prototyping and integration of algorithms: stereoscopic matching and bundle adjustment
- maintenance of algorithmic pipeline : adaptation and test of pipeline for data of different format
General Electric and Siléane, internships and end of studies projects.
General Electric : estimation of position of catheter of ablation, from a unique X-ray projection.
Sileane : point cloud registration to 3D primitives.