EN.601.783

Vision as Bayesian Inference

Tues/Thurs: 9:00-10:15 Spring 2025, Homewood Campus, Hodson 210 

Course Description

This course models vision as Bayesian Inference. It concentrates on visual tasks such as segmenting images, detecting objects in images, and recognizing objects. The course will also cover advanced topics including CNNs, Transformers, NERF, diffusion models, vision-language models and LLMs/ChatGPT. Its goal is to describe the state of the art techniques. The handouts consist of copies of the lecture notes and related papers.

Important information

• Instructor: Prof. Alan Yuille, ayuille1@jhu.edu.
• TAs: Junfei Xiao (junfei.xiao@jhu.edu), Qihao Liu (qliu45@jhu.edu)
• CAs: Haonan Zhang (hzhan240@jh.edu), Weina Dai (wdai11@jhu.edu)
• Office hours (OH):
  • Junfei Xiao: Friday, 4:00–5:00 PM (Zoom Link)
  • Qihao Liu: Tuesday, 8:00–9:00 PM (Zoom Link)
  • Haonan Zhang: Monday, 9:00–10:00 AM (Zoom Link)
  • Weina Dai: Monday, 1:00–2:00 PM (Zoom Link)
  Note: If no one joins within 30 minutes, TAs/CAs may leave early. In-person office hours are available by appointment. Please email us to schedule a time. Any changes will be posted in the "Discussions" section. 
• We encourages students  to initiate a discussion related to course in piazza.
• Grading Plan:
  • (tentative) 4 homework assignments (to be posted at Gradescope or Piazza) + bonus participation points. Late homework will not be accepted unless permission is obtained well in advance in documented extenuating circumstances. Please email Prof. Yuille (ayuille1@jhu.edu) for approval and cc TAs.
  • We encourage students to participate in class, by asking questions to the professor. TAs will periodically record the names of those who ask interesting questions and award them bonus points that contribute to their final grades (1 point per class session, with a maximum of 3 points). Students not asking questions will not experience any negative effects on their grades.

Homework

• Homework 1 is out on Gradescope. Deadline: February 25, 2025.
• Homework1 covers the first 3 weeks of the courses. It is released in the section "Assignments". Please submit directly in Gradescope.

Reading Material

• Course Notes and Handouts.
• Papers and Journal Articles.

Tentative Schedule

• More will be updated. Please see the schedule in Spring23 for reference. We will add more advanced topics including CNNs, Transformers, NERF, diffusion models, vision-language models and LLMs/ChatGPT.