Sign inSign up
Focusing Human Attention on the "Right" Visual Data thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Focusing Human Attention on the "Right" Visual Data

Published on Sep 18, 20128964 Views

Kristen Grauman

Widespread visual sensors and unprecedented connectivity have left us awash with visual data‐‐‐from online photo collections, home videos, news footage, medical images, or surveillance feeds. Which

ICME 2012 - Melbourne

Related categories

Computer VisionHuman Computer InteractionComputer Science

Chapter list

Focusing Human Attention on the “Right” Visual Data00:00
Visual data00:40
Automating visual processing01:15
"Semi-automating" visual processing (1)02:38
"Semi-automating" visual processing (2)03:15
The importance of data in recognition (1)04:48
The importance of data in recognition (2)05:56
Active learning for image annotation (1)07:25
Active learning for image annotation (2)08:30
Problem: "Sandbox" learning09:19
Our idea: Live active learning12:23
SVM margin criterion for active selection13:27
Sub-linear time active selection15:24
Background: Locality-Sensitive Hashing17:09
Hashing a hyperplane query (1)19:00
Hashing a hyperplane query (2)19:37
Hashing a hyperplane query (3)20:57
Hashing a hyperplane query (4)21:35
H-Hash result on Tiny Images (1)22:27
H-Hash result on Tiny Images (2)23:30
Live active learning25:21
Live active learning results (1)27:29
Live active learning results (2)28:55
PASCAL Visual Object Classes (VOC)29:34
PASCAL Live active learning results30:07
Summary so far31:00
"Semi-automating" visual processing32:00
Problem: Summarizing egocentric videos32:24
Potential applications of egocentric video summarization33:17
Existing approaches to video summarization34:46
Important person/object discovery35:30
Approach overview36:08
Data collection37:13
Crowdsourcing training data (1)38:02
Crowdsourcing training data (2)38:33
Learning region importance (1)39:02
Learning region importance (2)39:19
Learning region importance (3)39:41
Learning region importance (4)41:10
Segmenting the video into events42:06
Discovering an event’s key people/objects43:16
Generating a storyboard summary43:35
Results: Important region prediction (1)44:06
Results: Important region prediction (2)45:00
Results: Egocentric video summarization (1)47:50
Results: Egocentric video summarization (2)49:03
Results: Egocentric video summarization (3)49:12
Results: Egocentric video summarization (4)49:44
Summary51:06