Lecture 15 - Stanford Vision Lab
Lecture 15 - Stanford Vision Lab
Lecture 15 - Stanford Vision Lab
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Recognition<br />
1. Run part detectors exhaustively over image<br />
1<br />
1<br />
2<br />
3<br />
1<br />
2<br />
2<br />
1<br />
2<br />
3<br />
4<br />
3<br />
h<br />
0N<br />
<br />
0N<br />
0N<br />
<br />
0N<br />
1<br />
2<br />
3<br />
4<br />
<br />
<br />
<br />
<br />
<br />
<br />
e.g.<br />
h<br />
2<br />
<br />
3<br />
0<br />
<br />
2<br />
2. Try different combinations of detections in model<br />
- Allow detections to be missing (occlusion)<br />
3. Pick hypothesis which maximizes:<br />
p(<br />
Data | Object,<br />
Hyp)<br />
p(<br />
Data | Clutter,<br />
Hyp)<br />
4. If ratio is above threshold then, instance detected<br />
Fei-Fei Li<br />
<strong>Lecture</strong> <strong>15</strong> -<br />
66<br />
14‐Nov‐11