MeiChens Resume

1
Mei-Chens Resume
English-Language Presentation Practice

• Mei-Chen Yeh
• 08/03/2007

2
Scenario

• You are great interviewers!! Please be nice

3
Outline

• Education
• Work Experience
• Project
• QA

4
Education

• Pursuing Ph.D.
• Dept. of Electrical and Computer Engineering,
• University of California, Santa Barbara, CA
• M.S.
• Dept. of Computer Science,
• National Tsing Hua University, Taiwan
• Thesis Scalable Ideal-Segmented Chain Coding
• B.S.
• Dept. of Computer Science,
• National Tsing Hua University, Taiwan

5
Work Experience

• Research Assistant, Learning-Based Multimedia
  Lab, UCSB (2004 - present)
• Intern, Intel Corporation, Chandler (2006)
• Intern, Intel Corporation, San Jose (2005)
• RD Software Engineer, LITE-ON IT Corp., Taipei
  (-2004)

6
Outline

• Education
• Work Experience
• Project
• Face Annotation
• Machine Tagging for Photos
• Image Categorization
• QA

7
Face Annotation
Selected Project 1
8

• Provide an alternative way to organize photos
• Enable rapid text-based searches based on names

9
It is difficult for a computer to recognize
people!
Bush vs. Schwarzenegger
10
Serena Williams vs. Venus Williams
11
Moreover
Kobe Bryant is the only person I want to annotate!
12
Our Formulation

• Annotation
• Detection
• Recognition
• Name
• (Person of interest)
• Anonymous
• (Person dont care)
• Not a face
• (false positive)

13
Our Approach
Detected Face
Global Description (Pixel Intensities)
Local Description (Bag-of-Features)
Classifier

Unknown
Person 1
Person C
Non-Face
14
Status

• We achieved 80 recognition accuracy.

15
Machine Tagging for Photos
Selected project 2
16
Personal photos are not readily accessible in
large sets due to the lack of meta-data.
A large amount of photos (10000)
Find me the photos of Meis ELP presentation, if
any
?
17
Objective
Automatic creation of metadata!
textual query party
GUI
Off-line Process One photo might take seconds to
be processed
photos
Online Process Real-time photo search
photos
Database
metadata
Meaningful tags!
Meis Engine
18
Meis Engine
19
Online demo (http//lbmm.ece.ucsb.edu/php/photo/)
20
Image Categorization Using Text
Selected project 3
21
Classify each input image into one of the
predefined categories
Image Classifier
..
building
bus
highway
street
car
airplane
22
(No Transcript)
23
(No Transcript)
24
Our Framework
Input image
Categorization based on detected text lines
Categorization based on low-level cues
Fusion
Image category
25
Results
8.8 improvement
26
Summary

• Education
• Work Experience
• Project
• Experience in the fields of
• Pattern Recognition
• Machine Learning
• Multimedia Computing

27
Questions?

• Online demo of the tagging tool
  http//lbmm.ece.ucsb.edu/php/photo/
• Please check my website http//uweb.ucsb.edu/meic
  hen/ for project details, datasets, and more
  resources