Course Identification

• Mathematical background: Basic courses in Linear Algebra, Calculus and Probability. 

• Programming skills: All class assignments will be in Python. If you have a lot of programming experience but in a different language (e.g. C/C++/Matlab/Javascript) you will probably be fine.

• We also encourage students to take “Introduction to Computer Vision” course, which provides complimentary fundamental knowledge to ours. 

• Introduction & ML Basics

• Basics of Neural Networks

• Convolutional Neural Network

• Advanced Neural Networks Training

• Detection / Segmentation

• Visualizing and Understanding

• Sequences (Recurrent Neural Networks, Attention, Transformers)

• Generative Models (GANs, VAEs)

• Self-supervision in Images, Videos and Sound

• Learning from Videos

• CNN as representations and optimizers

• Implicit 3D representation 

• Graph Neural Networks 

• Deep Learning Theory

Upon successful completion of the course the students will be able to have:

• In-depth understanding of the fundamentals of deep-learning based methodologies in computer vision  (e.g., convolutional neural networks, optimization, back-propagation, generative image models)

• Gain knowledge about core deep learning algorithms, modern approaches and cutting-edge research for various visual tasks.

• Hands-on experience with deep learning for computer vision:

  • Implement neural networks and their components from scratch. 

  • Train, run and debug CNN models using leading frameworks (PyTorch) 

• https://www.deeplearningbook.org/

• Some of our basic topics are also covered in Stanford CS231n