en
0.25
0.5
0.75
1.25
1.5
1.75
2
Employing The Complete Face in AVSR to Recover from Facial Occlusions
Published on Nov 11, 20112903 Views
Existing Audio-Visual Speech Recognition (AVSR) systems visually focus intensely on a small region of the face, centred on the immediate mouth area. This is poor design for a variety reasons in real w
Related categories
Chapter list
Full face audiovisual speech recognition00:00
Overview - 100:03
Overview - 200:14
Visual models - 100:52
Overview - 301:17
Visual models - 201:38
Visual algortihms - 101:59
Visual algortihms - 202:20
Employed tropes02:46
Problems? - 103:13
Problems? - 203:14
Problems? - 303:22
Dissonance - 103:48
Dissonance - 204:04
Dissonance - 304:15
Berisha’s work04:33
Berisha et al05:21
Solutions? - 105:22
Solutions? - 206:32
Videos - 107:24
Videos - 207:32
Visual models - 308:00
Occlusions08:23
Formulation08:29
Testing09:04
Database09:23
Results - 109:37
Results - 209:57
Conclusions10:29
Future work10:48
References11:08