AV recognition  <br />Teacher: We*-ch*** L**<br />Presenter : P***A Hi**ng<br />
Outline <br />Motivation<br />Purpose scheme<br />Conclusion<br />Comment<br />
AV!OUT!70%successful rate<br />未來加強輪廓辨識,像是身形曲線、<br />器官等,才能更精準提升過濾功能。<br />Is this have more smart way to solve that?<br /...
Client<br />Server<br />Algorithm<br />pyNum<br />pySci<br />pyMatplot<br />Purpose scheme - System architecture<br />Trai...
Purpose scheme – Algorithm<br />Input<br />Regular<br />Section<br />Filter<br />FFT<br />二次微分<br />特徵比對<br />Sum up<br />
Eigen value <br />時域:音高+波形<br />頻域:共振峰<br />SOD<br />
For example <br />Mr. children – GIFT<br />Linked Park – New divide <br />Teriyaki boyz - Tokyo drift<br />
For example<br />Sector 01<br />Sector 02<br />Sector 03<br />Sector 04<br />Sector 05<br />
TokyoDrift <br />SOD<br />
Conclusion<br />Training Video in database by “D:”<br />Finish the processes of “sector” and “filter”<br />Releasing the A...
Questions<br />特徵值設計<br />分數評比的權重值<br />還有特殊案例<br />執行效率問題<br />成功機率<br />小發現:可以找出配樂喜好及主角分類<br />很憋得對角戲>搭配影像?<br />樣本夠大就可以...
Upcoming SlideShare
Loading in …5
×

Av Recognition

784 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
784
On SlideShare
0
From Embeds
0
Number of Embeds
3
Actions
Shares
0
Downloads
6
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Av Recognition

  1. 1. AV recognition <br />Teacher: We*-ch*** L**<br />Presenter : P***A Hi**ng<br />
  2. 2. Outline <br />Motivation<br />Purpose scheme<br />Conclusion<br />Comment<br />
  3. 3. AV!OUT!70%successful rate<br />未來加強輪廓辨識,像是身形曲線、<br />器官等,才能更精準提升過濾功能。<br />Is this have more smart way to solve that?<br />AV Killer讓Youtube笑了<br />
  4. 4. Client<br />Server<br />Algorithm<br />pyNum<br />pySci<br />pyMatplot<br />Purpose scheme - System architecture<br />Training<br />Process<br />SOD<br />DB<br />(MS SQL)<br />MIC<br />Identify <br />Process<br />AV’s<br />slice<br />
  5. 5. Purpose scheme – Algorithm<br />Input<br />Regular<br />Section<br />Filter<br />FFT<br />二次微分<br />特徵比對<br />Sum up<br />
  6. 6. Eigen value <br />時域:音高+波形<br />頻域:共振峰<br />SOD<br />
  7. 7. For example <br />Mr. children – GIFT<br />Linked Park – New divide <br />Teriyaki boyz - Tokyo drift<br />
  8. 8. For example<br />Sector 01<br />Sector 02<br />Sector 03<br />Sector 04<br />Sector 05<br />
  9. 9. TokyoDrift <br />SOD<br />
  10. 10. Conclusion<br />Training Video in database by “D:”<br />Finish the processes of “sector” and “filter”<br />Releasing the AV-killer version α<br />
  11. 11. Questions<br />特徵值設計<br />分數評比的權重值<br />還有特殊案例<br />執行效率問題<br />成功機率<br />小發現:可以找出配樂喜好及主角分類<br />很憋得對角戲>搭配影像?<br />樣本夠大就可以?<br />

×