Visual Learning and Recognition

12 units · Graduate

Spring: 12 units This graduate-level computer vision course explores representation and reasoning for large-scale data, such as images, videos, 3D data, and text, toward understanding the visual world surrounding us. Students will engage with a diverse selection of classic and recent research papers covering mid-level vision (grouping, segmentation), object and scene recognition, 3D scene understanding, action recognition, multimodal perception, vision-language models, multimodal deep generative models, efficient deep learning, and more. We will explore state-of-the-art neural architectures, including CNNs and transformers, and a wide range of supervised, semi-supervised, self-supervised, and unsupervised approaches for each topic above. Prerequisites: 10-701 Min. grade B or 16-385 Min. grade B or 16-722 Min. grade B or 15-781 Min. grade B or 16-720 Min. grade B Course Website: https://visual-learning.cs.cmu.edu/