2008年12月22日 星期一

[Meeting] 12.22

Interface:
- Load photos
- Classify by timing first
- Offer feature selection toolbar to classify the required category

2008年12月16日 星期二

[Meeting] 12.16

Current Progress:
SVM testing, Interface building, Feature extraction Revising.

Interface Specification:
1. Server:
- Database
- User, Password
- User, Class
- Class, PhotoPath
- Online
- Offline
- Users' individual folder
- Store photos in each folder
2. Client:
- Upload
- Download
- View
- Manual classification
- Personalized interface
- Group(?)

Classification Process:
1. Reaction Diagram














Next Meeting:
12/22 Mon.

2008年12月9日 星期二

[Meeting] 12.9

Today's work:
Setup the machine

Task:
1. Setup Matlab7.2 that can convert to library of .Net and test how to communicate with C#
2. Design CBIR function
3. Implement SVM by Matlab
- http://www.csie.ntu.edu.tw/~cjlin/libsvm/
- http://diary.blog.yam.com/yyblog/article/6193250
- http://luzhenbo.88uu.com.cn/Internet/libsvm_introduction_study.htm
- http://hi.baidu.com/obgnim/blog/item/41f39bdd44c73ee977c638c0.html
- http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf
4. Fetch metadata of photos
5. Store the features to database
6. Tune the weight of the fusion between content-based and metadata features
7. Design the interface

Specification:
1. Server system
- 2, 3, 6
- Communicate with SQL server
- Contains user management system

2. Client program
- Can upload photos to server

Next Meeting:
12/16 Tue.

2008年12月7日 星期日

[Meeting with Winston] 12.8

Methods of classification :
1. SVM

Features:
1. GPS
2. metadata+concept+time
3. EXIF-perl

2008年12月4日 星期四

[Meeting] 12.4

Today's work:
Determine the Automatical Classification Album be our topic of final project.

Schema:
1. Classify the photos in album.
2. Find the similar pictures in networking website.

Task:
1. Survey the two open source album systems
- http://www.phpalbum.net/
- http://gallery.menalto.com/
2. Survey the php framework
- http://www.codeigniter.com/
- http://www.php.net/manual/en/ (php manual)
3. Web Resources
- http://neural.cs.nthu.edu.tw/jang/
- Install http://doublekai.org/docs/CI_Install_Guide/CI.html
Questions:
1. Practical and creativity?
2. Communicate with PHP and Matlab?
3. How to retrieve the tag from photo?
4. Another interesting idea in album system?

Next meeting:
12/9 Tue.

2008年12月2日 星期二

[Meeting] 12.2

Today's work:
1. We've finished testing the face recognition of openCV. And it can identify the face of Lena successfully.

Schema:
1. Find the region of face from OpenCV (C++/Python)
- Find boundary of the face
- Normalize
- Map to low dimension
- Store location
2. Extract the features and classify pictures by SVM (Matlab/PHP/Python)
3. Store the features to database (MySQL)
4. Retrieve pictures form database to satisfy the query requirement (PHP/Python)
5. Display the result on the website (JavaScript)

Task:
1. Survey Python
- Tutorial http://www.freebsd.org.hk/html/python/tut_tw/tut.html
- Download http://www.python.org/download/
- 2, 3, 4, (5, 6, 9)
- Make document
- Useful Python framework
- Django http://www.djangoproject.com/
- TurboGears http://turbogears.org/
2. Survey OpenCV
- How to pre-process image

Questions:
1. Could a picture map to a location and store them before?
2. How to call Matlab from PHP or C# ?

Next meeting:
12/9 Tue.