Paper Reviews

Visual Recognition and Search

EECS 6890 Topics in Information Processing (3 Credits)

Instructors: Rogerio Feris, Liangliang Cao, and Jun Wang

   Home   |  Course Overview  |  Schedule  |  Resources  |  Projects  |  Presentations  |  Paper Reviews  |  Announcements

Paper Reviews


Each individual student (solo, no groups) will be required to write reviews of assigned papers (1-2 pages), addressing the following content:

  • Summary. Briefly summarize the paper and the key contributions using your own words.
  • Strengths and Weaknesses. Discuss the paper strengths and weaknesses. Try to identify limitations of the proposed method that were not mentioned in the paper.
  • Experimental Analysis. Comment on how convincing and comprehensive are the experimental results.
  • Proposed Extensions. Describe ways in which the work could be extended and comment on open research questions.

The reviews should be emailed to the three instructors. You can use up to 3 late days over the course of the semester. Beyond that, your submission will not be accepted.


Assigned Papers and Deadlines


March 11:  Perronnin et al, Improving the Fisher Kernel for Large-Scale Image Classification, ECCV 2010
You may check this IJCV paper to help you better understand the assigned paper above.


April 08:  Weinberger et al, Distance Metric Learning for Large Margin Nearest Neighbor Classification, NIPS 2005


April 29:  Shotton et al, Real-Time Human Pose Recognition in Parts from Single Depth Images, CVPR 2011