COL864: Special Topics in Artificial Intelligence
(Advance Computer Vision)
Even Semester 2018-19



Overview

Recent advances in algorithmic techniques, computation and memory technologies have reinvigorated interest in artificial intelligence (AI). Many of the successes in AI in last few years have come from its sub-area computer vision which deals with understanding, and extracting information from digital images and videos. This course will cover advanced topic in computer vision. We will examine data sources, features, and learning algorithms useful for understanding and manipulating visual data. The emphasis will be on scalability issues as well as acquiring the knowhow to work on interdisciplinary problems. The goal of this course is to give students the background and skills necessary to perform research in computer vision and its application domains such as robotics, healthcare, and computational photography. Students should understand the strengths and weaknesses of current approaches to research problems and identify interesting open questions and future research directions.

Course Content

  • Scene Understanding
    • Image Classification
    • Object Detection
    • Semantic Segmentation
  • Generating Images and Videos
    • Autoregressive Models
    • Variational Autoencoders (VAEs)
    • Generative Adversarial Networks (GANs)
  • Explainable AI
    • Interpretability and Explainability in Computer Vision Systems
    • Adversarial Attacks on Computer Vision Systems
  • Vision and Language
    • Image Captioning
    • Visual Question Answering (VQA)
    • Visual Dialog
There will be no textbook for the course and the teaching material will be based on the publicly available latest research papers and talks.

Course Policies

Prerequisites
  • There are no formal prerequisites for the course.
  • However, a student is expected to be familiar with machine learning techniques, especially deep neural networks, through any of the courses such as Machine Learning, Artificial Intelligence etc.
  • Basic knowledge about working with digital images and videos, through courses such as Computer Vision, Digital Image Processing may also be required.
  • Students interested in a brush-up may go through video lectures of some of these courses listed here.
  • The course is expected to be assignment heavy, and a student may be required to install and work with multiple deep learning libraries, at times through public web services. Proficiency in programming and working with multiple external libraries would be desired.
Marks Distribution
  • Mid Term Exam: 15 (There will be only one mid term instead of two minors)
  • Major Exam: 30
  • Assignments(3): 30
  • Course Project: 20
  • Attendance and Class Participation: 5
How to Fail in the Course
  • Scoring less than 50% marks in assignments or project head will lead to failing the course
  • Any plagiarism detected in any of the assignments/project will lead to failing the course
  • Other institute rules such as failing for skipping major exam will also apply
Other Policies
  • No deadline extension in any assignment or project submission
  • No auditing the course
  • Sit through permitted only for the PhD students

Lecture Slides


Acknowledgement

We would like to thank Microsoft for the USD 20K grant of Azure credits to be used by the students of the course.