ENEE698a: Pattern Recognition from Temporal Data

Prof. Rama Chellappa

Tuesday 4:00pm to 5:30pm
AVW 2120

So here are the outline for the projects for ENEE 698a. More information will be added as things get fleshed out, and I want to be very flexible as to what data is used in the project. If you have a dataset/topic that you would like to use please let me know, and we can talk about whether it would be suitable.
General Structure:

The project will be done in groups of 2.
Each group will select/be assigned a topic (see below for topics)
Code for each topic should be readily available (either Matlab or downloaded from the web), so evaluation (whether it is classification, segmentation, etc) of the method is what should be concentrated on.
Goal: Present a 15-20 minute presentation on what you discovered at the end of the semester.

Topics:

Dimensionality reduction : Look at linear (PCA/FDA) and non-linear reduction. Some examples are:
- Activity Recognition
- Dynamic Textures
- Traffic Analysis
- Surveillence
- Related Code: ISOMAP, LLE
- Data Sets: USF(1) USF(2) From the CMU MoBo database: Walking Slow Walking Fast Walking with a Ball See Dynamic Textures. Manifold Learning
Dynamic Textures : Look at different models of textures to see what is/is not captured in different models.
- Related Code: AR model, Model Distance
- Data Sets: Texture Database More Textures Even More
Tracking and Recognition
- Related Code: Mean Shift
- Data Sets: Fish More Fish Even More Fish Lots of Fish

The list of project partners and when they present are:

December 5th:
- Srikanth and Pavan
- Steve Tjoa and Avinash Varna
- Nima Mesgarani and Daniel Garcia-Romero
- Ling Liu and Yao Li
- Ruonan Li and Nick Horner
December 12th:
- Vikranjit Mitra and Punyaslok Purkayastha
- Xiaodong Yu and Yi Li
- Zhe Lin and Guanyu Zhu
- Aniruddha Khembavi and Kaushik Mitra
- Soma Biswas and Mohamed Abdelkader