FR
EN
Computer VisionDécision (GSI)
5ECTS (EU)
1Coefficient
16:00Hours
Séquencement
Learning objectives
• Know the great tasks of computer vision (detection, recognition, segmentation, matching) • Study in detail two different applications: matching images with points of interest, and segmentation
Prerequisites
| Module | Teaching unit | Semester |
|---|---|---|
| Object-oriented programming C++ | Mathematics and Computer Science 1 | S5 |
| Advanced Object-Oriented Programming | Engineering Sciences | S8 |
| Image Processing | Analysis | S9 |
Program
• Overview of the main tasks of computer vision: detection, recognition, segmentation, matching • Detection of points of interest for image matching ( coins detector, multi-scale representations, SIFT points) • Graph cut image segmentation: random Markov fields, maximum flow problem, Ford-Fulkerson method, graph cut Programming is done in C++ with the OpenCV library
Assessment
Two TPs evaluated: 50%, 50%
References
Computer vision: models, learning and reference. S.J. Prince, Cambridge University Press, 2012.
Skills
4| Ref. | Verb | Description | Level |
|---|---|---|---|
| C2_1 | analyze | Master the main differential operators on images (gradient, hessian, Laplacien) and understand their roles in spot detection | 2 |
| C2_1 | analyze | Be able to construct a multi-scale representation of an image and detect local extrema that will serve as points of interest | 2 |
| C2_1 | analyze | Understanding the role of conventional terms (data attachment, regularization) in an optimization problem for image processing, and in particular for segmentation | 2 |
| C2_1 | analyze | Master one of the possible transformations of an image into a graph, maximize a flow over a graph, and extract a segmentation from it | 2 |