1st
Title: Graduate Student Grooming Habits: Automatic Annotation
Date: Thursday, April 1st, 2010
Time: 4:01 p.m.
Place: GHC 5404
A central problem for 21st Century science will be the analysis and understanding of the graduate student grooming patterns. My talk will be concerned with topics within this area, in particular automatic annotation of security camera footage from several campuses. I will discuss a comprehensive facial expression, hairstyle, and attire identification pipeline and storage database we have built. This has enabled use to identify a vocabulary of >10K “messemes”; the unique vocabulary or disheveledness. Using this vocabulary we can analyze the distribution with respect to age, gender, field of study, and cycle of the moon. By traversing several months of footage we were able to find a strong correlation with the progression of a semester and the number
of consecutive days students wear the same pair of jeans.
I will try to inter-relate these studies with similar work using hidden electronic chemical detectors (“e-nose bugs”), which enable one to comprehensively probe the frequency of student bathing. At the end I will bring these together to propose a general theory showing statistically significant correlations between the frequency of stains on T-shirts, students’ hormonal cycles, and conference deadline extensions.
Throughout I will try to introduce some of the computational algorithms and approaches that are required for such grooming pattern annotation — i.e. the construction of annotation pipelines, developing algorithms allowing us to identify patterns without having to actually look at too many grumpy, unshaven faces, and refining approaches for scoring wardrobe color mismatching.
Breathmints will be served, and there will be a limited supply of plastic combs compliments of several hotel chains.
