Course Information
- Lecturer Dr. Frank Lenzen , Prof. Björn Ommer
- Lecture extend (2+2 SWS)
- Time & date:
- Lecture Mo 14:15-15:45
- Exercise: Mo 16:00-17:30,
- Room : HCI (Speyerer Straße 6)
- Lecture: large seminar room 2.floor room H2.22
- Exercise : small seminar room 3rd floor 3. OG
- Language: English
Contents
The lecture covers two topics:
Topic 1: Deep Learning, in particular Deep Convolutional Neuronal Networks (CNNs). As basics for Deep Neural Networks we will discuss convolutions, filters , Fourier analysis, wavelets overcomplete bases, learning, optimization, before going in to detail on neural networks.
Topic 2: Multiview geometry and 3D scene estimation. Subtopics are camera models and camera geometry, stereo, structure from motion, optical flow, depth estimation
Misc
- standard certificates ('Schein') after passing exercises and oral exam
- certificate for attendance ('Sitzschein'): regular attendance required (absence in not more than 2 lectures )
- Exercises partly build on MATLAB. Alternatively the students may use octave oder NumPy.
Material
Additional material is posted in moodle.
Since currently not all students have moodle accounts, <a href="https://hcicloud.iwr.uni-heidelberg.de/index.php/s/hpG8zwopXHB4TBP"> this </a> is an alternative link:
(same password as for moodle).