Label Embedding for Text Recognition thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Label Embedding for Text Recognition

Published on Apr 03, 20142532 Views

The standard approach to recognizing text in images consists in first classifying local image regions into candidate characters and then combining them with high-level word models such as conditional

Related categories

Chapter list

Label embedding for text recognition00:00
Table of Contents - 100:11
Table of Contents - 200:17
Problem: Text recognition in natural images - 100:19
Problem: Text recognition in natural images - 200:56
Problem: Text recognition in natural images - 301:15
Related work - 101:30
Desired framework03:14
Table of Contents - 303:34
Model formulation - 103:38
Model formulation - 204:50
Model formulation - 305:26
Q1. Image embedding05:51
Q2. Label embedding 06:15
Q3. Learning the W in the compatibility function - 107:38
Q3. Learning the W in the compatibility function - 208:20
Table of Contents - 409:22
Experiments on License Plate Recognition - 109:29
Experiments on License Plate Recognition - 210:28
Experiments on Scene Text Recognition12:13
Correctly recognized words13:01
Failure cases - 113:13
Failure cases - 213:18
Failure cases - 313:42
Conclusions14:12
Thanks15:11
Q3. Learning the W in the compatibility function - 315:35
Related work - 217:14