NAVAL TRAFFIC ESTIMATE IN VENICE
This Computer Vision project estimates the naval traffic in Venice from a webcam video of the Grand Canal. We make use of motion detection algorithms such as adaptive background subtraction, then morphological filters and blob detection are implemented to count the number of boats and estimating the traffic (integral average, median and flow rate).
This Visual Data project exploits image classification trough on a Visual Search System based on a hierarchical K-means tree structure. The features for the images are represented by SIFT keypoint descriptors. It is based on Bag of Visual Words theory and the score system we used for the object retrieval is the TF-IDF (term frequency inverse document frequency) score.
Football players detection and tracking
This Computer Vision project, developed as Bachelor Thesis within an internship at Xtensa Srl in Trento, exploits football players detection and tracking trough Yolo V3. It aims at understand if deep learning based tracking can outperform classic object tracking algorihtm as Mean-shift.
Image Matching: SIFT & SURF
This Computer Vision project investigates the results of two popular scale-invariant feature detectors, SIFT and SURF, to find features in images and matching similar images.
This Computer Graphics project shows a demo of the Solar System, navigable through keyboard and with .jpg textures on the planets.
This Multimedia Data Security project investigates some techniques to detect if an image is tampered and localizate the inserted artifact. We used PRNU-based or JPEG-based methods.
