In the table below, in the page numbers pp. xx/yy, xx refers to the page number in the hard copy of Szeliski's Computer Vision: Algorithms and Applications, and yy refers to the electronic version.1
This schedule is based on the plans for the last offering of the class. During that class, we discussed some additional topics (such as homomorphic filters) which are not included here, but we may discuss again during this offering. Some changes I would like to make this time around are to focus more on the machine learning aspects of computer vision and to focus more on creating features from pixels. So we may deviate significantly from the schedule, especially later in the quarter.
| Week | Day | Topics | Reading | Lab |
|---|---|---|---|---|
| 1 |
1 | Overview of Modern Computer Vision
Image representation and manipulation: Pixels, color channels, and filtering |
1.1*, 1.2, 1.3*
3.1.1-3.1.3 (p. 91/) |
Lab 1: Simple Image Manipulations |
| 2 | Filtering: edge conditions | 3.1.4, Eq. 3.41 (p. 112/128), 3.2 (p. 98-107/111-122) | ||
| 2 |
1 | Filtering: differentiation
Summarizing an image: Histograms and Blocks 2D homographic point transforms Changing pixels: Thresholding |
Appendix A
2.1-2.1.2 (pp. 29-36/) |
Lab 2: Introduction to filtering and histograms |
| 2 | Review of linear algebra
Filtering: Hessian (Hessian not covered) Homographic image warping. Interpolation. |
6.1 - 6.1.4 (pp. 275-282/), 3.5.1 | ||
| 3 |
1 | Multi-resolution blob detection and image pyramids
Machine Learning and Features |
4.1 (p. 183-184/), A.1.2 (p. 647-649/)
3.5 (all), 4.1.1, Sub-section Scale Invariance (p. 191-193/) |
Lab 3: Warping images |
| 2 | The SIFT feature-point description | 4.1.2 | ||
| 4 |
1 | Margin | TBA | |
| 2 | Composite image transforms | |||
| 5 |
1 | Composite image transforms -- exercise | Lab 5: Multi-scale interest-point detection (SIFT)
Feedback Survey | |
| 2 | Half Exam 1 Testing machine-learning algorithms: Performance metrics, ROC curves |
4.1.3 (p. 200-204/) | ||
| 6 |
1 | RANSAC -- Random Sampling and Consensus | 6.1.4 | Lab 6: Image stitching by SIFT/SURF feature matching |
| 2 | 3D transforms: Rotation, Translation, Camera projection | 2.1.4-2.1.5 (pp. 36-52/) | ||
| 7 |
1 | 3D transforms: Focal lengths, Vanishing points and Camera calibration | 6.3 (pp. 288-295/) | Lab 7: 3D Camera Geometry and Calibration |
| 2 | Half Exam 2 Radial distortion |
2.1.6 (pp. 52-54/) | ||
| 8 |
1 | Survey: Face Recognition | 14.2.2-14.2.3, papers | Lab 8: Preliminary Implementation of a Computer Vision Project |
| 2 | Survey: Face Recognition | 14.2.2-14.2.3, papers | ||
| 9 |
1 | Surveys: Structure from motion and 3D reconstruction and 3D Reconstruction | 7 (all) and 12 (all) | Lab 8: Completion of Implementation of a Computer Vision Project and documentation, and report writing |
| 2 | Survey: 3D Reconstruction | 12 (all) | ||
| 10 |
1 | Survey: Structured Light (Kinect) | Presentation of Computer Vision Projects | |
| 2 | Review | |||
| 11 |
TBA Final Exam | |||
1 It is rare that I supply both... if you would like to complete the table, please feel free to save this page, insert the missing numbers, and email the page to me. I will edit this page from time to time, so please contact me if you plan to do this over more than a day.