COMP 6820 - AI and Advanced Computer Vision 3

This course provides a rigorous, hands-on exploration of the foundational principles and state-of-the-art architectures driving modern computer vision. Moving beyond introductory image classification, the course deeply investigates contemporary deep learning models necessary for high-level visual understanding and synthesis. Students will begin by mastering core deep learning concepts, including advanced Convolutional Neural Networks (CNNs), optimization techniques, and the critical practice of Transfer Learning. The curriculum then transitions to the core advanced tasks: implementing and benchmarking sophisticated systems for Object Detection (e.g., YOLO, Faster R-CNN), Semantic and Instance Segmentation (e.g., Mask R-CNN), and an introduction to 3D and Pose Estimation. In the latter half of the course, emphasis shifts to emerging frontiers, including the mechanics of Attention Mechanisms and Vision Transformers (ViT), and the implementation of Generative Models such as GANs and Diffusion Models for image synthesis and novel content creation. The course culminates in an exploration of practical deployment challenges (model efficiency, bias, and ethics) and a substantial capstone project where students design and implement a complete solution to a real-world advanced computer vision problem. Prerequisite(s): COMP 5480 Artificial Intelligence. Hybrid