Computer Vision

Semester:
SS 2018
Type:
Lecture
Lecturer:
Credits:
V3 + Ü1 (6 ECTS credits)
Upcoming Dates
# Date Time Room Type Title
1 10:15 - 11:45 Audimax Gr Lecture Local Features I
2 10:15 - 11:45 Audimax Gr Exercise Exercise 3
3 14:15 - 15:45 Audimax Gr Lecture Local Features II

Lecture Description

Cameras and images form an ever-growing part of our daily lives. Billions of images and massive amounts of video data are becoming available on the Internet. Large search engines are being created to make sense out of this data. And more and more commercial applications are coming up, e.g. in surveillance and security, on consumer devices, for video special effects, in mobile robotics and automotive contexts, and for medical image processing. All those applications are building on visual capabilities. For us humans, those capabilities are natural. But how do we actually accomplish them? And how can we teach a machine to perform similar tasks for us?

The goal of Computer Vision is to develop methods that enable a machine to "understand" or analyze images and videos. This lecture will teach the fundamental Computer Vision techniques that underlie such capabilities. In addition, it will show current research developments and how they are applied to solve real-world tasks. The lecture is accompanied by Matlab-based exercises that will allow you to collect hands-on experience with the algorithms introduced in the lecture (there will be one exercise sheet roughly every two weeks).

Literature

In the last decades, Computer Vision has evolved into a rapidly growing field with research going into so many directions that no single book can cover them all. We will mainly make use of the following books: Forsyth &: Ponce: Computer Vision - A Modern Approach

  • D. Forsyth, J. Ponce, Computer Vision - A Modern Approach, Prentice Hall, 2002
  • R. Szeliski, Computer Vision - Algorithms and Applications, Springer, 2010
  • R. Hartley, A. Zisserman, Multiple View Geometry in Computer Vision, 2nd Edition, Cambridge University Press, 2004

However, a good part of the material presented in this class is the result of very recent research, so it hasn't found its way into textbooks yet. Wherever research papers are necessary for a deeper understanding, we will make them available on this web page.

Course Schedule
Date Title Content Material
Introduction Why vision? Applications, Challenges, Image Formation
Image Processing I Linear Filters, Gaussian Smoothing, Multi-scale Representations
Exercise 1 Intro Matlab
Image Processing II Image Derivatives, Edge detection, Canny
Structure Extraction Line Fitting, Hough Transform, Gen. Hough Transform
Exercise 2 Derivatives, Edges, Hough Transform
Segmentation I Segmentation as Clustering, k-means, EM, Mean-Shift Segmentation
Segmentation II Segmentation as Energy Minimization, Markov Random Fields, Graph Cuts
Recognition I Sliding Window-based Object Detection, Chamfer Matching, HOG, AdaBoost
Local Features I Interest points, Harris Detector, Hessian Detector, Scale Invariance, Local Descriptors, SIFT
Exercise 3 Mean-shift segmentation, Sliding-Window Detection
Local Features II Specific Object Recognition with Local Features, Geometric Verification, RANSAC
Deep Learning I Intro Neural Networks, Backpropagation, etc.
Deep Learning II CNNs, Current architectures, VGGNet, GoogLeNet, ResNet
Deep Learning III CNNs for object detection
Exercise 4 Interest Point Detection & Matching, Homography Estimation
Deep Learning IV CNNs for semantic segmentation, human pose estimation
Deep Learning V CNNs for matching, video processing
3D Reconstruction I Multi-view Stereo Basics, Disparity, Triangulation, Epipolar Geometry, Essential Matrix, Correspondence Search
Exercise 5 CNNs
3D Reconstruction II Camera Parameters, Calibration, Triangulation, DLT
3D Reconstruction III Fundamental Matrix, Eight-Point Algorithm, Active Stereo, Outlook to SfM
Repetition -
Exercise 6 Eight-point algorithm, RANSAC, Triangulation
Disclaimer Home Visual Computing institute RWTH Aachen University