Visual Relationship Detection with Language Priors thumbnail
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Visual Relationship Detection with Language Priors

Published on Oct 24, 20164335 Views

Visual relationships capture a wide variety of interactions between pairs of objects in images (e.g. "man riding bicycle" and "man pushing bicycle"). Consequently, the set of possible relationships is

Related categories

Chapter list

Visual Relationship Detection with Language Priors00:00
Images/100:08
Images/200:31
Example/100:43
Example/200:53
Problem formulation/101:12
Problem formulation/201:20
Problem formulation/301:24
Problem formulation/401:34
Problem formulation/501:43
Related work/101:49
Related work/202:26
Visual Genome dataset02:35
Observation 102:49
Observation #2/103:29
Observation #2/203:37
Observation #2/303:43
Visual module/Language module/104:00
Visual module/Language module/204:16
Visual module/Language module/304:21
Visual module/Language module/404:26
Visual module/Language module/504:42
Visual module/Language module/605:01
Visual module/Language module/705:15
Visual module/Language module/805:33
Visual module/Language module/905:45
Visual module/Language module/1005:54
Quadratic explosion06:26
Long tail distribution06:37
Training the visual module/107:09
Training the visual module/207:22
Training the visual module/307:28
Training the visual module/407:34
Training the language module/107:44
Training the language module/208:05
Training the language module/308:19
Training the language module/408:26
Training both modules iteratively08:36
Our results/108:45
Our results/208:53
Our results/309:03
Our results/409:07
Our results/509:18
Our results/609:25
Our results/709:34
Ablation study/109:44
Ablation study/209:57
Ablation study/310:09
Ablation study/410:19
Example/110:32
Example/210:42
Zero shot detection/111:01
Zero shot detection/211:26
Zero shot detection/311:32
Zero shot detection/411:42
Thank you!11:49