Suggestions for CSE 523/524 Projects

 

 
 
 

Here are some sample topic areas, that could stimulate your choice of a narrower, more specific project. Please talk to me about your ideas so that you can get help in focusing your work. Topics 1-4 were added Jan 2005
 

  1. Dynamic Probabilistic Networks for Video analysis. The project would be to implement ideas contained in the following papers, and test them either on video of multiple activities, or on fMRI data (3D “movies” of the brain): Gong, S. and T. Xiang. Recognition of group activities using dynamic probabilistic networks. in IEEE International Conference on Computer Vision. 2003. or  Vogler, C. and D. Metaxas. Parallel hidden markov models for American sign language recognition. in ICCV. 1999.
  2. Face Shape and Motion Recovery: We have multiple research projects on this area. You will work together with senior PhD students.
    1. You could work on the new generation hardware and software forcapturing human facial expression. Please talk to me for details.
    2. Implement: "Recovering Shape and Reflectance Model of Non-Lambertian Ojects from Multiple Views", Tianli Yu, Ning Xu and Narendra Ahuja, CVPR04
    3. Implement: "Real-Time Combined 2D+3D Active Appearance Models", J. Xiao, Simon Baker, L. Matthews and T.Kanade, CVPR04
    4. Facial expression analysis and transfer using multinear models. You would need to assist in the capture and analysis of data.
  3. Generalized PCA (applied to fMRI data)
  4. Object Recognition and local features:. Implement and compare the methods described in the following papers:
    1. Object class recognition using discriminative local features
    2. Weak Hypotheses and Boosting for Generic Object Detection and Recognition
    3. Scale and affine invariant interest point detectors
  5. Juggling Tutor
  6.  Virtual building Project, joint work with Olaf Hall Holt. A number of projects are available, talk to me for more details.
  7. Tracking objects using "condensation". Tracking objects in cluttered backgrounds. More details.
  8. Bundle Adjustment: A method that recovers both camera and scene geometry. Brief description. Implement this method to recover the camera geometries and a few feature points on sets of face images. You would need to search the web to find a generic face model first.
  9. Comparison of face recognition techniques. You can implement at least three techniques and compare them. You can start from this paper and the face recognition page.
  10. Implement Illumination-insensitive Face Recognition Using Symmetric Shape-from-Shading
  11. Hidden Markov Models and Gesture Recognition.    Hidden Markov Models have been succesfully applied to speech recognition. Here we will explore their applicability to gesture recognition.
  12. Vision and Learning. Learning object models of non-rigid objects, for example. Learning image-based representations. Active learning of shape or x. Implement Support Vector Machines for face detection and recognition.