1. Midterm Total points: 75 Total points graded: 36 Median: 31/36
Mean: 28.23/36
4 students have full marks (36) thus far
25 students have >90%*36 marks thus far

2. Final project: Each diving in-depth to a sub-area of computer vision
Read a research paper
Present it twice
First round: 3 mins (presentation) + 2 mins (taking questions)
Second round: 5 mins (presentation) + 3 mins (taking questions)
Write a report (Due: 11/22/16, 11:59PM)
>= 1 page
Letter page (8 ½ x 11 in), 1 inch margin, single spacing, 12 point Times New Roman
40% of overall grade

3. Logistics Find your paper and when to present here:
Paper Assignments:
Papers:

4. Logistics
Upload slides to webcourses before the class (Important)

5. How to read research papers? (1)
Ask questions before you start
What are the motivations for this paper?
What problem is being solved?
What is the proposed solution?
What experiments are designed to test the solution?
What are the evaluation methods and metrics
What are the contributions?
What are the future directions?
How is the paper related to what you previously knew?
How is the paper related to other works?

6. How to read research papers? (2)
Repeat: decreasing marginal gain until convergence?
Get the general idea first
What problem is being solved?
What are the main steps in the solution?
What is the high-level idea of the solution?
Draw analogies to what you know well
Read in order:
Abstract, conclusion, problem statement if it exists, intro, approach, experiments
Divide and conquer
How to read research papers by Dr. Mubarak Shah:
The three-pass approach by Dr. Srinivasan Keshav:

7. How to present research papers (in class)?
Title, authors (full name), authors’ institutes, your name and
Motivation of the research (1 slide)
Problem statement (1 slide)
What is being solved?
Why is it an important problem?
Main contributions of the paper (1 slide)
Studied a new and important problem
Proposed a novel approach
Improved or extended existing methods
Compared several popular methods
Explored a variety of use cases (many datasets of different kinds)
Presented new theories
Introduced new methodologies or tools to computer vision

8. How to present research papers? (3)
Title, authors (full name), authors’ institutes, your name and
Motivation of the research (1 slide)
Problem statement (1 slide)
Main contributions of the paper (1 slide)
Some details of the proposed approach
Experiments
Data, features, baselines, evaluation metrics, results