Special Topics in Data Science, DS-GA 3001.005/.006
Logistics
-
DS-GA 3001.005 (Lecture)
Tuesdays 2pm-3:40pm 60FA room 110 -
DS-GA 3001.006 (Lab)
Tuesdays 3:50pm-4:40pm 60FA room 110 -
Office hours: Thursdays 2:00pm-4:00pm 60FA room 606
Instructors
Jean Ponce (jean.ponce@inria.fr)
Matthew Trager (matthew.trager@cims.nyu.edu)
TAs:
Jiachen Zhu (jiachen.zhu@nyu.edu)
Sahar Siddiqui (ss12414@nyu.edu)
Grading
Four programming assignments (60% of the grade) + final project (40% of the grade). Assignments should be submitted using the NYU class site.
- Excercise 1 on camera calibration (zip file). Due on October 1st.
- Excercise 2 on Canny edge detector (zip file). Due on October 22nd.
- Excercise 3 on mean shift (zip file). Due on November 12th.
- Excercise 4 on neural networks (ipynb notebook). Due on December 17th.
- Final project: list of suggested papers is available here. Send an email to Matthew to validate a project. Poster presentations will be on December 17th at 1.30 pm in the 7th floor open space. You should also submit a 1-2 page report by the same date.
Collaboration policy: You can discuss the assignments and final projects with other students in the class. Discussions are encouraged and are an essential component of the academic environment. However, each student has to work out their assignment alone (including any coding, experiments, and derivations) and submit their own report/notebook.
Syllabus
- Part I: Low level Computer Vision
- Filters, edge detection, visual features.
- Radiometry, shading and color.
- Part II: 3D reconstruction
- Camera models, one-view geometry.
- Multi-view geometry, stereo, SFM.
- Part III: Recognition
- CNNs for object detection and semantic segmentation.
References:
- D.A. Forsyth and J. Ponce, “Computer Vision: A Modern Approach”, second edition, Pearson, 2011.
- R. Szeliski, “Computer Vision: Algorithms and Applications”. (PDF)
- R. Hartley and A. Zisserman, “Multiple View Geometry in Computer Vision”, Cambridge University Press, 2004.
Lectures
Week | Date | Topic | Instructor | Slides |
---|---|---|---|---|
1 | 9/3 | Course overview, image formation | JP | Slides |
2 | 9/10 | Camera geometry and calibration I | JP | Slides |
3 | 9/17 | Camera geometry and calibration II | JP | Slides |
4 | 9/24 | Linear and nonlinear filters, edge detection | MT | Slides |
5 | 10/1 | Interest points (Harris, SIFT), robust estimation (RANSAC, Hough transform) | MT | Slides |
6 | 10/8 | Radiometry and color | JP | Slides |
10/15 | No class | |||
7 | 10/22 | Texture and image segmentation | MT | Slides |
8 | 10/29 | MRF and graph cuts, single-view geometry | MT | Slides |
9 | 11/05 | Epipolar geometry | JP | Slides |
10 | 11/12 | Stereo | JP | Slides |
11 | 11/19 | Structure from motion | JP | Slides |
12 | 11/26 | Intro to recognition, ML, and deep learning | MT | Slides |
13 | 12/03 | Neural networks, object classification and detection | MT | Slides |
14 | 12/10 | Weakly supervised and unsupervised methods | JP | Slides |
Acknowledgements
Much of the material for this course relies on the Computer Vision course given at ENS Paris by Mathieu Aubry, Karteek Alahari, Ivan Laptev, and Josef Sivic. Many of the slides are taken from James Hays, Svetlana Lazebnik, and Derek Hoeim.