In addition to the extra questions on some of the assignments and on the final exam, I am asking the 505 students to write a short research paper on a course topic, essentially a critical literature survey.

The goal is to read two papers: a recent high quality research paper and its source paper, ideally a very well known high-impact paper.

There are several ways to approach this topic. You could select a recent paper, say from CVPR 2016, ECCV 2016, or ICCV 2015. Both should be available inside the UBC VPN. Your task will be to read the paper and determine the foundational method that the paper extends. Then read both papers, describe both of them, showing how the recent paper builds upon the original.

Another way is to look for a high-impact paper older paper in your area of interest and then find a recent paper from the conferences or other recent conferences that cites your older paper.

For example, the Alexnet paper "ImageNet Classification with Deep Convolutional Neural Networks" is enormously influential. I found it by Googling "Alexnet convolution" and got to the page "The 9 Deep Learning Papers You Need To Know About", not necessarily an authoritative source be reasonably well structured. Many papers (over 10,000) cite this paper.

I am open to suggestions you might have about other ways to approach this task. I can discuss the papers with you but the process of identifying the research paper and its predecessor will be valuable for you.

Another source of important papers: the Koenderink prize winners (including SIFT) are given at ECCV for fundamental contributions in computer vision that stood the test of time. Or the Longuet-Higgins Prize fro CVPR.

Best papers at any of CVPR, ICCV, and ECCV are good choices, too, for the recent paper - then you work backwards.

The research paper (5-10 pages) is due at the end of the exam period, April 28. You should send me a proposal for your two papers by email by April 7, giving you 3 weeks to read the papers and write your report.