Prof. Lorenzo Baraldi

• Slides dai docenti, pubblicate su Moodle - Paper di riferimento, segnalati sulle slide, e in generale pubblicati in CVPR; NIPS; ICCV; TPAMI - Testo di riferimento: D. Fosith, Computer Vision

Argomenti del programma: 1. Introduction: Computer Vision and Artificial Intelligence; history; the current research trends; the market, applications, future trends. 2. Data manipulation in PyTorch: tensors, strided representation, tensor operations. Computational graph construction and automatic differentiation. 3. Image processing: image, video, resolution; histograms; color and other spaces; compositing and matting; OpenCV labs.

Argomenti del programma: 1. Hardware and architectural foundations for large-scale AI CPU vs GPU processing flows Multi-GPU architectures Multi-Node architectures Going parallel: data, model, pipeline, tensor parallelism 2. State-of-the-art Deep Learning architectures for Computer Vision, Natural Language Processing e Multimedia The attention operator, self-/cross- attention and its variants The Transformer architecture Transformer-based…