Sylvain GOURGEON

– Leading the innovation for the company's computer vision capabilities. – Evaluate, benchmark, and integrate state-of-the-art deep learning methods (classification, object detection, segmentation) to enhance product performance and reliability. Fine-tune neural networks by adapting hyperparameters and dataset (data augmentation). – Combine deep learning with traditional image processing to deliver hybrid production-grade computer vision pipelines adapted to industrial constraints. – Explore and validate low-cost embedded Al technologies, including NPU-accelerated neural network inference on Raspberry Pi. – Develop new functionalities for Optical Coherence Tomography (OCT) and X-ray tomography systems using advanced algorithms and model-based approaches. – Lead intellectual property activities on deep learning: analysis of competitors' patents, innovation scouting, and drafting patents (e.g., deep learning-based multimodal, multi view inspection patent — WO2025003618A1). – Act as a technical reference for computer vision topics, supporting engineering teams with guidance, design reviews, and best practices. – Work cross-functionally with software, hardware, and optical engineering teams to ensure seamless integration of Al modules into products.

– Developed real-time algorithms and embedded vision solutions for industrial inspection systems. – Designed and deployed real-time image processing algorithms in C/C++ for embedded platforms. – Built an OCR system for mold identification, enabling automated and reliable tracking in industrial workflows. – Created an autonomous image registration algorithm for automated machine configuration based on recurrent pattern. – Applied background subtraction for real-time anomaly detection. – Developed and deployed deep learning classifiers/detectors, improving detection/decision accuracy across multiple products. – Used optical modeling techniques to design custom algorithms for enhanced image analysis. – Integrated Python modules into C/C++ production software, enabling hybrid workflows and dynamic runtime configuration. – Collaborated closely with hardware and domain experts to optimize embedded performance. – Supervised interns and contributed to knowledge sharing across the R&D team.

– Developed a C/C++ pattern-matching algorithm with rotation invariance and noise robustness. – Optimized performance and validated accuracy across multiple datasets.

– Documented methodology and presented results to the research team. – Implemented the Unicode standard in a Natural Language Generation software (C/C++)

– Implemented C/C++ lightweight cipher algorithms on an MSP430 to make a benchmark. – Publication: Survey and benchmark of lightweight block ciphers for MSP430 16-bit microcontroller.