Project Report

• Introduction


• Motivation


• 用 social network 的角度分析人的重要性，以一種 Layout 呈現，且讓 user 更方便的 browsing



• Related Work


• 鼎翔: PhotoAgent


• Layout: Treemap, tag cloud



• Problem Definition


• 資料(photo)


• 取得的 metadata 有哪些 (person, event, time, place, ...)


• 分析的是什麼 (person, 算 weight)


• 希望達到的效果 (PowerPoint 的 constraint)



• Proposed Approach


• Social Network Analysis


• PhoEver Picture


• Time Sequence (*取個名字)



• Evaluation


• 設計 2~3 個 tasks，比較各種照片/SN 呈現方式的優缺(適用時機)



• Discussion and Future Work


• Discussion:
  1. time sequence 可以加入更多考量
  2. 與 evaluation 相關 (? 等 evaluation 有結果再寫)


• Future work
  1. layout optimization
  2. interesting finding
  3. help annotation




• Conclusion (optional)

[PhoEver] UI

If there is any questions, please tell me. ^______^
Step 1:

Step 2:

Step 3:

Step 4:

[PhoEver] Photo Display UI

Here is our photo display UI.

[PhoEver] Photo Display

Ordered Layout:


Random Layout:

[Discussion] Evaluation

User study: compare traditional photo display and our photo display
1. User Satisfacation
- investigate users' preference (traditional v.s. our)

2. Accuracy
- design some missions and let users finish it
- evaluate how ratio users match or miss the ground truth

Project Idea

Team Members: Janet,Evelyn, Emily
From: Intelligent Agents Lab, NTU CSIE
Project Idea:
Following on previous work in our lab, PhotoSlap and PhotoAgent, we want to use the photo's metadata to build a personal social network. Base on the social network, we want to use a novel way to visualize the different weight (familarity or importance) of different people for a user. In visualization, we may use the treemap to display photos. If the mouse hover on one photo, it maybe show all related photos in a specific sequence. (time sequence, familiarity sequence...etc.)

Multimedia Analysis and Indexing

Multimedia Analysis and Indexing
2007 Spring NTU CSIE
Lecturer: Winston Hsu
TA: Tien, Min Chun