Amer Saidi

Development and deployment of Machine Learning solutions for real-time object detection

• • Development, training, and evaluation of Machine Learning/Deep learning models for object detection.

• • Development of signal and image processing algorithms for computer vision applications.

• • Optimization and deployment of ONNX models for real-time inference on x86 architectures and embedded systems (ARM64).

• • Design of real-time perception pipelines in C++ and Python with ROS2 and GStreamer.

• • Deployment of an SSD300 model on NVIDIA embedded platforms, including Jetson TX2 and Clara AGX Xavier.

Machine Learning

• • Embedded Systems

◦ • Software Engineering Design of a RISC-V processor on FPGA dedicated to linear regression

• • VHDL Design: design and specification of a single-cycle RISC-V RV32I processor on Kintex-7 FPGA.

• • C Development: optimization of a linear regression algorithm under memory and hardware precision constraints.

• • Integration and validation: configuration of the RISC-V compilation chain, program compilation, data transfer via UART, and validation of the solution on the target platform.

• • Computer Vision

◦ • Software Engineering Development of an industrial vision application for quality control

• • Development of an image processing application for automated inspection and quality control.

• • Design of computer vision algorithms for defect detection in manufacturing.

• • Evaluation and validation of the solution's performance in an industrial environment.