CS 281b/ECE 281b: Advanced Topics in Computer Vision
B. S. Manjunath, manj at ece. Rm 3157, Engineering I. Tel: 805.893.7112
Office hours: T 11:15am-12:30 pm, or by appointment (or drop by and see if I'm available)
Teaching Assistant: Pratim Ghosh. Office hours: Thursdays 3-5pm, PHELPS 1435.
Meeting Times and Locations
Lecture: M/W 10:00am - 11:50 am, PHELPS 1431
About the course: (see course outline)
This is a graduate level course in computer vision covering selected topics in computer vision. Initial lectures will cover fundamental topics in computer vision, and these are complemented by recent papers. Topics include: image formation, shape from shading, camera calibration and multiview geometry, optic flow and structure from motion, segmentation and tracking, feature extraction with applications to face detection, recognition, human activity analysis, visual search and retrieval.Discussions will be based on selected articles (recently published journal and conference articles). Grading will be based on class participation (10%), home works and quiz (25%), paper presentations (15%), final project (report+presentation) 50%.
Update: Project presentations will be scheduled during the final week of the quarter June 2 and June 4 (Friday).
Suggested Project Themes: Detecting and blurring faces in videos, tracking and counting bikes in a camera network data, human activity analysis in aerial images. The UCSB camera network will be used to collect the data for the projects. It is expected that the students work in pairs on the projects as well as their paper presentations.
March 31: Introduction (slides). UCSB Bike Path Camera Network (slides). HW #1 is due on Monday, April 5, in class. Today's lecture was based on a paper on Gestal theory and computer vision by Desolneux, Moisan and Morel (paper).
April 5: Image formation basics (slides). HW #2 is due on Monday, April 12, in class. Additional reading (bookchapter section). Radiometry (slides).
April 7: Shape from shading (slides). Additional notes on recovering shape from shading. Also, see the handout (distributed in class) - Vic Nalwa Chapter 5.
April 9: Feature extraction basics (sldes-corner, slides-sift).
April 12: Edge detection and segmentation - Part I (slides). HW #3 (due April 19).
April 14: Optic flow (slides, bookchapter). HW #4 (due April 26) [image data]
April 16: Structure from X and Multiview geometry (X= stereo, motion) (slides1, slides2)
April 19: Multiview geometry contd. - camera models (slides)
April 21: Tracking (slides)
Paper presentation schedule is now online. Camera network book chapter for presentation is also available.
April 26: RANSAC and application to computer vision (slides)
April 28: Image Segmentation (slides)
April 30: Student projects- brief presentations of project plans.
May 3: Paper discussions.
A. B. Chan, Z. S. J. Liang, and N. Vasconcelos. “Privacy preserving crowd monitoring: counting people without people models or tracking”, in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2008. [Min-Chi Shih, Yi-Chu Wang, Lakshmanan Nataraj, Sevada Abraamyan]. (critique)
Yun Zhai, Rogerio Feris, Lisa Brown, Russell Bobbitt, Arun Hampapur, Sharath Pankanti, Quanfu Fan, Akira Yanagawa, "Composite Event Detection in Multi-Camera and Multi-Sensor Surveillance Networks," Book chapter. [Norma Saiph Savage, Victor Fragoso, Domagoj Bari?evi?, Diana Delibaltov, Marco Rodriguez-Suarez, Paul Filitchkin] (critique) [this may get updated on May 6, 2010. Check the web page for an update.]
May 10: No class
May 12: Jim Kleban will give an overview of search and retrieval in large image/video databases and also highlight some of his work on the SpirtTagger system. (note: he will be giving his thesis presentation on Thursday, May 6.) (slides)
May 17: Segmentation (curve evolution methods, contd.) (slides)
May 19: Face representation, detection and recognition. (slides)
May 24: No Class
May 26: TBD
May 31: Memorial day holiday
FINAL PROJECT REPORTS: DUE by 5PM on Monday, June 7.
There is no required format for the reports. Organize it similar to a short conference paper, with an abstract, problem statement, a brief review of previous work and what you did this quarter for the project. Make sure that you clearly emphasize those aspects that you consider interesting/important in your project. You can also include any pictures/graphics/etc that highlight your overall project.
June 2: Project presentations (15 mins + 5 mins Q&A)
W01: 10:00 - 10:20AM, Vignesh Jagadeesh <email@example.com> Karthikeyan Shanmuga Vadivel <firstname.lastname@example.org>
W02: 10:20 - 10:40AM, Russ McLoughlin <email@example.com>, Ritesh Lala <firstname.lastname@example.org>
W03: 10:40 - 11:00AM, Lakshmanan Nataraj <email@example.com>, Sevada Abraamyan <firstname.lastname@example.org>
W04: 11:00 - 11:20AM, Domagoj Bari?evi? <email@example.com>, Diana Delibaltov <firstname.lastname@example.org>
W05: 11:20 - 11:40AM, Xin Yang <email@example.com>, JieJun Xu <firstname.lastname@example.org>
June 4: we will start at 3PM and continue till all projects are presented.We will have food for dinner!
F01: 3:00 - 3:20PM, Harinandan S <email@example.com>, Santosh Kumar C S <firstname.lastname@example.org>
F02: 3:20 - 3:40PM, Min-Chi Shih <email@example.com>, Yi-Chu Wang <firstname.lastname@example.org>
F03: 3:40 - 4:00PM, Carlos Torres <email@example.com>, Rodrigo Perez <firstname.lastname@example.org>
F04: 4:00 - 4:20PM, Dilip Gopalakrishna <email@example.com>, Sreenidhi Anand <firstname.lastname@example.org>
F05: 4:20 - 4:40PM, Marco Rodriguez-Suarez <email@example.com>, Paul Filitchkin <firstname.lastname@example.org>
Break: 4:40 - 5:00PM
F05: 5:00 - 5:20PM, Carter De Leo <email@example.com> Zefeng Ni <firstname.lastname@example.org>
F06: 5:20 - 5:40PM, Norma Saiph Savage <email@example.com>, Victor Fragoso <firstname.lastname@example.org>
F08: 5:40 - 6:00PM, Wendy H. Chun <email@example.com>, Aruna Jammalamadaka <firstname.lastname@example.org>
Solutions: HW2, HW3, HW4