- 10am-12pm Wednesdays (online)
- Read Student FAQs
5 ECTS module corresponds to about 100-125 hours of study time. This includes the several hours of lectures (30+ for undersgrads, 20+ for postgrads). The academic year structure is dense and good time management is essential.
Contacting the lecturer
A Discussion Forum is/will be opened online (e.g. with Blackboard or/or Teams/Yammer provided to students in TCD) to post queries and is used as the primary tool for discussion on topics related to the course. Emails from students in my taught modules might be overlooked for logistic and time management reasons. If so, please ask questions during classes or flag this email to me directly at the end of the class.
For the past decade, neural networks have become very important for computer vision. Convolutional Neural Networks (CNNs) have integrated many processing steps from older computer vision techniques ("vintage" computer vision). This course aims at covering CNNs as well as several older techniques that are all relevant today for computer vision.
Keywords: Image processing, feature detection and matching, image registration, recognition and segmentation - Motion flow and object tracking in video - Mathematics for computer vision - Machine/Deep Learning for computer vision
Last year module webpage CS7GV1 (MT2019)
This module will be assessed by a combination of various continuous assessments including quizzes, in-class test, and projects.
There are a lot of books in the Library that are relevant to the course. Online references and ressources are also available e.g.:
- Computer Vision: Algorithms and Applications, Richard Szeliski, Springer 2010 - see also 2020 version of this book
- Computer Vision: Models, Learning, and Inference , Simon J.D. Prince, Cambridge University Press 2012
- Computer Vision: A modern approach, Forsyth and Ponce, Pearson 2012 (available in TCD library)
- Programming Computer Vision with Python, Jan Erik Solem
- Deep Learning, I. Goodfellow et al., MIT 2016 press
- CVonline: Vision Related Books
- Numerical Tours of Data Science
- Computer Vision on Twitter
A popular language currently used is Python, but Matlab (available in TCD) or Octave works well as well for rapid testing of techniques. See the following video https://youtu.be/oXPX8GIOiU4 for the recent information on Deep Learning Software.
Vintage Computer vision
I got really into a bunch of 1980s-era papers about histogram thresholding, and wrote a weird paper. I sent it to ECCV assuming the reviewer response would be "why are you writing a direct response to two 40 year old papers" but they loved it, so hey. https://t.co/BgZ9dETc8p— Jon Barron (@jon_barron) July 24, 2020
Computer vision with deep learning
- Read Energy and Policy Considerations for Deep Learning in NLP, E Strubell et al. ACL 2019.
- See Stanford CS class CS231n (Spring 2017) Youtube playlist and website
Videos for my Fall 2019 course "Deep Learning for Computer Vision" are now on YouTube!— Justin Johnson (@jcjohnss) August 10, 2020
This is an evolution of @cs231n that I used to teach at Stanford:
- All content refreshed
- New topics: Transformers, Video, 3D, etc
- HW in @PyTorch + @GoogleColab https://t.co/6nqZKTpmxv