Daily Working Activities
I made this blog to have a track of my research working, especially for myself, and to whom it may concern.
Wednesday, February 09, 2005
3 - 10 February 2005
Web Crawler
- Dealing with multithreading crawler
- Fish algorithm vs Shark algorithm
Wednesday, February 02, 2005
2 February 2005:
Investigate Multidimensional Access Method:
- Feature vector: Data object or metadata? (I gues, it is metadata)
- Bounding sphere/hypersphere, hypercube etc for multidimensional (n>3 data)
- Implementing SH Tree for 3D Similarity?
- CLAPACK http://www.netlib.org/clapack/CLAPACK3-Windows.zip
- Matrix Template Library http://www.osl.iu.edu/research/mtl/
CLAPACK is the result of converting "line by line" from Fortran to C :(. Impossible to understand form the C code.
Monday, January 31, 2005
29-31 January 2005:
Face Recognition System:
- Debugging original source code and modifying it (Matlab source code) in accordance with theoretical procedure
- Modifying FRS in MFC
- K-Means
24 - 28 January 2005:
Interoperable Workspace on WWW
- Searching & Learning SMTP dan POP3 Library: Blat222, Pop3 Monitor, MimeSniffer, PJNSMPT, STunnel
- Building Simple Interoperable File Storage using gmail.com (function module: loading file)
Sunday, January 23, 2005
17-23 January 2005
Paper writing:
- Writing paper "Self-Generating Similarity Evaluation"
- Investigate an eigenface algorithm -> there is a gap between theoritical procedure and source code
Sunday, January 16, 2005
16 January 2005
Eigenface Face Recogition
- Building face recognition using available eigenface database
- We use eigenface approach to extract the feature vector
- MFC and Matlab. I investigate that in this stage there is no need to store data in any RDMS'
Wednesday, January 12, 2005
Face Recognition (Determine the dirrection of research):
- Multiple feature face recognition or data-content based rule of classification? Still not clear!
- Searching the library for processing image to get eigenface
Monday, January 10, 2005
Face Recognition (Evaluate the feature vector):
- I come to the conclusion that left eye-right eye-nose, is too simple as a feature. Allthough there are some sites implementing such method, there is no enough information in Internet about the detail technique, and how is the result if there are so many faces
- Trying to move to other method: eigenface: investigate how it works
- Finding some eigenface database, converting to appropriate format