Skander Oueslati

Tasks: • Maintaining/Developing the reader/scheduler of the CNN graphs for the renesas R-CAR platforms. • Maintaining/Developing the RVX framework (Renesas OpenVX implementation) for the different renesas R-CAR platforms. • Implementing the unsupported CNN layers at the RVX layer (example: constant_padding, y_padding). Key Words: C/C++, Caffe, Tensorflow, CNN, R-CAR, OpenVX, Linux.

Tasks : Vision part for the FollowMe robot based on the nvidia jetson tx2 board. • Design and implement a real-time image-matching algorithm based on the Colour Histogram. • Design and implement a real-time tracker to follow a specific person. Key Words: C++, OpenCV, OpenPose, Nvidia, Nvidia jetson tx2, Linux.

Tasks : • Handling the 3D tracker for the side camera project. • Velocities estimation of tracked vehicles. • Parameterizing the Kalman filter used for the 3D tracker. • HeadLights detection and tracking • Creating OpenVX nodes for different targets (ARM, DSP) in the TDA4XSoC, using the TI Deep Learning framework. • Refactoring Vehicle front and side Camera demos. • Designing and developing an automated tests framework for the camera code (MobiTest).

Computer Science

Preparation for Engineering Studies : Mathematics, Physics, Engineering Science, etc.