After successful completion of the course, students are able to name, explain and contrast the most important theories, principles, concepts and algorithms of computer vision. Their knowledge and understanding corresponds to the state-of-the-art literature in the field of computer vision.

The lecture will cover advanced computer vision methods in depth:

• Introduction to CV
  • What is Computer Vision?
  • Why is Vision hard?
  • State-Of-The-Art Applications
• Image Formation
  • Pinhole Model
  • 2D Transformations
  • 3D Transformations
  • Camera Matrix
  • Photometry (What is Light?)
  • Thin Lens Camera
• Image Filtering
  • Point Operations
  • Linear Filters
  • Image Pyramids
• Local Features
  • Corner Detection
  • SIFT
• Image Classification
  • Challenges of Image Classification
  • Classification Methods (KNN, Clustering, Decision Trees, Random Forest, Linear Classification)
  • Neural Networks
  • Convolutional Neural Network (CNN)
  • CNN Architectures
  • Pretraining and Finetuning
• Object Detection
• Image Segmentation
• Instance Segmentation
• 3D Reconstruction
  • Epipolar Geometry
  • Structure from Motion
  • Bundle Adjustment
  • Stereo Matching
  • Block Matching
• 3D Data
  • Image-based Neural Networks
  • Voxel-based Neural Networks
  • Point-based Neural Networks
  • Graph-based Neural Networks
  • Coordinate-based Neural Networks

Lectures of the theoretical concepts by means of slides, and programming tasks.

ECTS breakdown (estimate, recommendation):


28 hrs. lecture
54 hrs. preparation units, individual preparation and tests
68 hrs. solving exercises and hand-in meetings
---------------------------------------------
150 hrs. corresponds to 6 ECTS at 25 hrs. each.

Submission of 5 exercises

Taking 1 test

Mathematics: vector and matrix calculus, from linear algebra
Programming, object-oriented programming
Computer Vision knowledge, from module Introduction to Visual Computing (both parts) and its prerequisite modules.

Python knowledge advantageous, but not required.