Lecture 11 Problems

(I won’t be posting solutions for this problem set because of the nature of the problems.)

Problem 1

We’re going to find out what the latest and best techniques are for a specific computer vision problem: real-time object tracking.

Part a. Using an LLM of your choice, ask for a list of the best methods currently known for real-time object tracking. In your writeup, show me the prompt you used, tell me which LLM model you used, and give me the names of at least three methods that the LLM recommended. For each one, do enough background research to tell me:

• The name of the method
• If the method really exists (i.e., was it an LLM hallucination?)
• In which year was the method published?

Part b. Let’s do this again with Google Scholar. Run a search for “real-time object tracking”. Find three papers with these properties:

• They were published within the last three years
• The title sounds on-topic for our purposes
• They have a relatively high number of citations

(Of course, remember that it’s normal for a paper published three years ago to have several times more citations than a paper published this year.)

For each of these three papers, download the paper and open it. Looking only at the figures, try to screenshot an image that shows what this paper can do.

In your writeup, give me:

• The name of the paper
• The screenshot showing what this paper does

Part c. Once more! This time we’ll use Papers With Code. Under “Browse State-of-the-Art > Computer Vision > Video”, I see two tags related to Object Tracking: “Multiple Object Tracking” and “Video Object Tracking”. I’d like you to pick the three promising papers, based on recency, popularity, and relevance.

Include these in your writeup:

• The name of the paper
• A screenshot showing what this paper does
• A one-sentence justification as to why you picked this paper
• A link to a GitHub repo for code for this paper

Part d. Based on all of the above research, which paper would you pick if you needed to do real-time object tracking? (It’s okay to give a qualified / hedged answer.)

Problem 2

Computer vision research gets published at conferences. Most conferences are yearly events consisting of oral talks (for the best papers) and poster presentations (for everyone else). These are generally considered the best conferences in computer vision:

• CVPR - Computer Vision and Pattern Recognition
• ICCV - The International Conference on Computer Vision
• ECCV - The European Conference on Computer Vision

Some computer vision papers get published at top conferences in adjacent fields:

• NeurIPS - Conference on Neural Information Processing Systems
• ICML - International Conference on Machine Learning
• ICLR - International Conference on Learning Representations
• SIGGRAPH - ACM Conference on Computer Graphics

For this problem, find a list of accepted papers from a recent (past, not future!) computer vision conference. Answer these questions in your writeup:

• Part a. Scan the list of papers. What are three interesting computer vision problems that you see people working on?
• Part b. Who’s paying for all this? Who are the top sponsors? Do you see any surprises? What did it cost to attend this conference?
• Part c. Find a paper that won a “Best Paper” award at this conference. Include a screenshot showing what it can do, and write one or two sentences summarizing what this paper is about.