Moctar Mounirou

- Contribute in designing ,training and enlightening deep learning models focused on image restoration and quality enhancement... The efforts enable efficient deployment of high-performance models directly within GoPro’s camera systems, bringing AI-powered enhancements to users in real-time without external processing. - develop specialized image processing algorithms tailored for GoPro’s camera products, focusing on shading calibration, stitching for 360 degree cameras etc.

• Image processing algorithms for Automatic License Plate Recognition (ALPR) : optimized and simplified 1-D algorithms for license plates localization and characters segmentation. • design and training of CNN for the Optical Character Recognition (OCR) • TinyML : enlightening and optimization of neural network architectures for the OCR to be embedded in ALPR cameras. • C/C++ implementation and optimization of the inference of the CNN performing OCR to be embedded in an old camera equipped with a C6x type DSP • Implementation and integration of image processing and/or machine learning algorithms for: ✓ real-time autofocus ✓ Real-time anomalies detection on ALPR cameras (tampering, blur, etc.) that can impact camera performances. • C/C++, Python, OpenCV, Visual Studio, SV

** Fast, Automatic and Robust approaches for computing Pseudo Geological ages from 2D and 3D seismic images based on regularized energy functionals minimization** - Generic Formulation of Pseudo-geological ages estimation as an inverse problem; - Proposal of an energy functional that respects the basic principles of geology, takes seismic unconformities and seismic faults into account; - Evaluation of various regularization methods (Total Variation, Generalized Total Variation) and penalties (L1-Norm and L2-Norm) that allows to introduce various prior information such as geometric deposit models or geological constraints coming from geologists; - Convex Optimization problems, Proximal algorithms; **Automatic reconstruction of seismic horizons ** - Surface deformation by "active contours"; - Solving Large linear systems with Direct solvers (Cholesky decomposition, LDL, CG, etc.), iterative (Gauss Seidel, Jacobi, etc.) Or using "Multigrid" approaches . - Matlab, Java

**automatic segmentation of interpreted seismic images - active contours; - Voronoï diagrams; -Java programming; - Matlab

Automatic reading of images of scanned documents by analyzing and segmentating colors (chromatic and achromatic) . - Image processing; - C ++ programming in Visual Studio; - OpenCV; - Matlab