Mustafa b.

Specialization in Computer Vision and Machine Learning under the supervision of Dr. Hedva Spitzer. Thesis focused on ultrasound image enhancement and segmentation using machine learning and classical image processing techniques. Developed and implemented deep learning models for image segmentation, object detection, super-resolution, HDR, video processing, image colorization, and image generation across various computer vision applications.

Expertise in optical systems and computational imaging. Final Project: Developed a Differential Interference Contrast (DIC) microscopy system for quantitative phase estimation, enhancing imaging precision and analysis under the supervision of Prof. Natan Shaked. Published a research paper in The Royal Society Interface Focus Journal (2019). Developed an object detection pipeline for small datasets using transfer learning, optimizing pre-trained models to achieve high accuracy in data-constrained environments. Designed and implemented an object segmentation model for highly imbalanced medical data (liver tumors) using a U-Net architecture combined with a voting-score ensemble technique to improve sensitivity and robustness. Built an end-to-end image colorization framework by integrating Generative Adversarial Networks (GANs) with a KNN-based dictionary memory module to enhance color consistency and contextual accuracy. Implemented a human activity recognition system using a hybrid CNN-LSTM architecture, enabling temporal feature extraction from sensor data for improved sequence classification performance.