Syllabus
The class will roughly be divided into thirds. The first third will focus on "bottom-up" methods for object detection and classification. The second third will focus on "top-down" 3D pose recovery. The final third will focus on exploring "advanced" methods applied to the project.
January 21: Introduction
Introduction to the problem. Introduction to the object dataset. Course organization. Reading assignments.
Introduction [pdf]
Slides on generating 3D object models [pdf]
Mundy Slides on Geometric Object Recognition [pdf]
Data and support code in /course/cs296-4/2009/scansOrtho/
Assignment 1 (due in class Jan 28): In Matlab make a 3x4 "contact sheet" of "training images" of one of the 3D objects using the 3D models stored above (eg. use subplot). Make sure to put your name on it.
Class mailing list: csci2950-q
Useful code for manipulating meshes: http://meshlab.sourceforge.net/
January 28:
Readings (see reading list for full citations):
1. Mundy
2. Dickinson
3. Sapp et al.You are to hand in 1 page summarizing the papers, giving your thoughts about them, contrasting them, etc. Your summaries are due in class on the day of the readings.
Updated: Datset design and description of assignment 2 (due in class Feb 4) [pdf]
Overview of other datasets from Fei-Fei/Fergus/Torallba [pdf]
Class exercise: data collection. Bring a camera to class.
February 4:
Overview of modern object recognition methods.
Hand in: contact sheets for assignment 2 and summay of readings.
Readings:
D. Lowe. Distinctive image features from scale-invariant keypoints.
International Journal of Computer Vision, 60(2):91–110, 2004. [pdf] [code]
Histograms of Oriented Gradients for Human Detection, by N.Dalal, B.Triggs. CVPR 2005 [pdf] [demo video] [software] [PASCAL datasets] [Dalal code] [PHOG code]
See also the PASCAL site and other links on the readings page.
Leibe and Grauman sliding window slides [pdf]
Example objects/segmentations, readings, Assignment 3 [pdf]
February 11:
At this point we have some initial images and initial ground truth. The next step is to try implementing a standard detector based on HOG features.
Readings
Fergus, R. , Perona, P. and Zisserman, A., "Object Class Recognition by Unsupervised Scale-Invariant Learning", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2003) [pdf]
S. Belongie, J. Malik, and J. Puzicha. Shape matching and object recognition using shape contexts. IEEE Transactions on Pattern Analysis and Machine Intelligence, 24(4):509–522, 2002. [pdf] [web] [code]
February 18:
Slides [pdf]
Readings:
P. Felzenszwalb, D. McAllester, D. Ramanan
A Discriminatively Trained, Multiscale, Deformable Part Model
Proceedings of the IEEE CVPR 2008 [pdf][webpage]
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
Svetlana Lazebnik, Cordelia Schmid, Jean Ponce, CVPR 2006 [pdf]
Possibly useful: Grauman's pyramid match code
Februrary 25:
Everyone reads one of the four papers. Pre-assigned individuals or groups will present the papers (roughly 15-20 minute presentations and discussions).
People not presenting should make more progress on the detector, send me some preliminary results, and be ready to talk about them.
Readings:
Visual learning and recognition of 3-D objects from appearance
International Journal of Computer Vision, Volume 14 , Issue 1 (January 1995)
Hiroshi Murase, Shree K. Nayar [pdf] (presentation: Andres and Dan)Joerg Liebelt, Cordelia Schmid, Klaus Schertler, "Viewpoint-independent object class detection using 3d feature maps", CVPR 2008 [pdf] (presentation: Jadrian)
Pingkun Yan, Saad M. Khan, and Mubarak Shah, 3D Model based Object Class Detection in An Arbitrary View, IEEE International Conference on Computer Vision (ICCV), 2007. [pdf] (presentation: Peng)
March 4:
Readings (read one and send me your summary):
A. Thomas, V. Ferrari, B. Leibe, T. Tuytelaars, B. Schiele, and L. Van Gool. Towards multi-view object class detection. In Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2006. [pdf] (presentation: Tim)
Bastian Leibe, Ales Leonardis, and Bernt Schiele, "Combined Object Categorization and Segmentation with an Implicit Shape Model", In ECCV'04 Workshop on Statistical Learning in Computer Vision, Prague, May 2004. [pdf] [code] [video1] [video2] (presentation: Luke)
Fidler and A. Leonardis. Towards Scalable Representations of Object Categories: Learning a Hierarchy of Parts. CVPR 2007. [pdf] (presentation: Jung)
March 11:
Reading:
Using Specularities for Recognition
Margarita Osadchy David Jacobs Ravi Ramamoorthihttp://www.cs.berkeley.edu/~ravir/specularities.pdf
Background
Here's a paper that is not about vision but about environement mapping for graphics
Environment Matting and Compositing
Douglas E. Zongker, Dawn M. Werner, Brian Curless, David H. Salesin
http://grail.cs.washington.edu/projects/envmatte/
but the techniques might be useful to capture properties of the glass for recognition.
March 18:
Reading:
3D Pose Refinement from Reflections
Pascal Lagger, Mathieu Salzmann, Vincent Lepetit, Pascal Fuahttp://cvlab.epfl.ch/publications/publications/2008/Fua08.pdf
project proposals due -- come prepared to stand up and describe your project. Feel free to send me visuals in advance if that helps.
March 25: Spring Break, no class
April 1:
April 8:
April 15: Putting it all together. Results and demos.
April 22: No class, work on projects