Welcome to the course website for “Multimedia Information Retrieval” 2011/2012!
This lecture gives an introduction to models and algorithms underlying modern multimedia search technology.
The focus will be on retrieving image and video content (though other media like text and audio will be touched as well).
- Lecturers: Dr. Adrian Ulges, Prof. Thomas Breuel
- Credits: 2C+1R, 4 credit points
- Language: English
- Level: Master (entry level)
- Requirements
- Introduction to Pattern Recognition (recommended, but not required)
- Introduction to Image Processing and Image Understanding (recommended, but not required)
- basic probability theory and analysis
- Literature (available in computer science library):
- Duda, Hart, Stork: Pattern Classification, 2nd edition
- Gonzalez, Woods: Digital Image Processing
- Bishop: Pattern Recognition and Machine Learning
- Exam: written or oral (to be discussed in class)
- More information: see KIS / Modulhandbuch
News
- Studying for exams
- As the exams draw closer, I just wanted to point you to some literature that you might find helpful
- “Pattern Classification” (Duda, Hart, Stork) and “Pattern Recognition and Machine Learning” (Bishop) are the standard reference books when it comes to pattern recognition (classification, clustering, and dimensionality reduction).
- Your co-student Sebastian Palacio (thanks Sebastian) pointed out this on-line book on the same topic: "Bayesian Reasoning and Machine Learning" (Barber). I don't know it but Sebastian found it quite helpful.
- When it comes to image analysis issues, you definitely want to check out “Digital Image Processing” (Gonzalez, Woods).
- Jan 13: EXAMS
- Exams will be taken in oral form by Prof. Breuel and Adrian Ulges
- We have reserved Apr 12 and Apr 13 for the exams
- To register for your exam, please contact Ingrid Romani (secretary at iupr dot com). Ingrid will coordinate the time slots for the exams (you will have to register with her first and then also with the “Prüfungsamt”).
- Jan 13: There will be no lecture on Jan 19. We will have two lectures on Jan 26 instead, one at 11:45 (usual time and place) and one in the afternoon (time and place like for the tutorials).
- Dec 03: Exercise sheet no. 4 is on-line - sorry for the slight delay. The sheet is to be submitted as usual in the tutorial next Thursday. In case of any questions, feel free to contact Adrian or Damian.
- Nov 25: Wan-Lei has provided an updated set of slides for his lecture yesterday (no. 6) - you find them below. If anything about the lecture content is unclear (the topic yesterday was maybe the most challenging one in the whole course, at least if you're not familiar with image processing), feel free to contact Wanlei or me with questions.
- Nov 17: As discussed in today's lecture, tutorials next week will be held on *Wednesday, Nov 23, 13:45-15:15 @ DFKI, room: 2.04*.
- Nov 17: A slightly updated version of the slides for lecture 4 has been uploaded (correcting typos and 1-2 index errors in formulas).
- Nov 15: Exercise sheet no. 3 is on-line. We'll have to shift the exercises next week (i.e., it is not clear yet when you have to submit your solutions). This will be discussed in this Thursday's lecture.
- Oct 26: Slides for lecture 02 (supervised learning) are up. In general, we will try to upload slides Wednesday night.
- Oct 20: The time slot for the exercises is settled to room 48-210, Thursday 15:30-17:00. The first exercise on October 27th will cover the first exercise sheet (basic maths)
- Oct 20: Lecture starts today!
Slots
- Lecture: Thursday, 11:45-13:15, 46-260
- Tutorials: Thursday, 15:30-17:00, 48-210
Contact
- Adrian Ulges
- adrian dot ulges at dfki dot de
- DFKI - room 2.06
- +49-631-20575-419
- Wanlei Zhao
- zhao at iupr dot com
- IUPR - room 48-456
Schedule
Lecture schedule:
| block | date | lecturer | content | details | slides & demos | exercise sheets & materials |
|---|---|---|---|---|---|---|
| 1 | 20.10.11 | Adrian Ulges | introduction | course formalities, definitions, history, lecture outline, introduction | slides | sheet |
| 2 | 27.10.11 | “ | pattern recognition (supervised) | basic decision theory, classification, classifier combination | slides R script | |
| 3 | 03.11.11 | ” | pattern recognition (unsupervised) | clustering, dimensionality reduction | slides | sheet material NumPy Website NumPy Tutorial |
| 4 | 10.11.11 | “ | text retrieval | document retrieval, retrieval models, relevance feedback, data structures | slides | |
| 5 | 17.11.11 | ” | image representation I | definitions, features and their properties, invariance | slides | sheet material |
| 6 | 24.11.11 | Wan-Lei Zhao | image representation II | local features: interest point detection, patch description, matching | slides slides(updated) | |
| 7 | 01.12.11 | Adrian Ulges | image representation III | local features vs. global features recap, color-based features | slides backup slides | sheet material |
| 8 | 08.12.11 | “ | similarity search I | applications, prototypical architecture, distance measures, object search | slides | |
| 9 | 15.12.11 | ” | similarity search II | scalability: kd-trees, locality-sensitive hashing | slides python script | |
| 10 | 05.01.12 | “ | visual recognition I | applications, challenges, face detection, face recognition, object recognition, visual words | slides | sheet material |
| 11 | 12.01.12 | ” | visual recognition II | object category recognition - SVMs and large-scale nearest neighbor | slides | |
| 12 | (19.01.12) | “ | visual recognition III | concept detection and concept-based video retrieval, concept selection, research | slides | sheet material |
| 13 | 26.01.12 | ” | video representation | shot boundary detection, keyframe extraction, motion estimation, motion-based applications | slides | |
| 14 | 02.02.12 | Damian Borth | audio and social signals | audio features and social signals, categorization, recommender systems, collaborative filtering | slides |
