• Developed a counterfeit coin detection project with a 98% accuracy , using a microscope to collect data/images of real and fake coins, and implemented a statistical edge-based feature selection method, based on a research paper (DOI: 10.1007/s11042-020-09447-8), The project involved feature extraction, PCA, and bag of words to reduce data dimensionality than training a classifier on the extracted features.• Created a people detection and tracking system using YOLOv5 and DeepSort, connected to a MongoDB database and hosted on a remote server, utilizing security cameras to detect the number of people entering and exiting a building.• Optimized face detection and face recognition models through POT and NNCF from OpenVINO, resulting in a doubling of FPS on the same device.