en-de
en-es
en-fr
en-sl
en
en-zh
0.25
0.5
0.75
1.25
1.5
1.75
2
Image and Video Tagging in the Internet Era
Published on Jul 18, 20115934 Views
Automatically converting visual content into textual description has long been a dream of a number of multimedia, computer vision and machine learning researchers. Image and video tagging, which has b
Related categories
Chapter list
Image and Video Tagging in the Internet Era00:00
Learning To Tag00:07
About Xian-Sheng Hua00:12
Why Need Tag Recommendation?00:32
Outline - 101:49
Tag Distribution on Flickr01:50
Outline - 202:34
What's Happening02:38
Tag Recommendation03:17
Straightforward Method03:57
Top Five Sites04:39
Characteristics of Internet Multimedia05:12
Learning to Tag05:41
Variety of Internet Multimedia Applications06:37
Tag Content Correlation06:54
Image - Conditioned Tag Correlation07:10
One of the Key Challenge. Search07:25
Learning To Tag07:57
Evolution of Media Tagging08:27
Results - 108:48
Results - 209:25
Results - 310:03
Different Types of Tags10:56
Flickr Distance11:06
Multimedia Information Retrieval12:12
Synonyms12:17
Image Similarity/Distance - 112:18
Outline - 312:47
Image Similarity/Distance - 212:48
Outline - 412:53
Learning Based Tagging12:56
Image Similarity/Distance - 312:59
Image Similarity/Distance - 413:03
Concept Similarity/Distance 13:06
WordNet Distance13:18
Learning-Based Tagging/Annotation13:28
Google Distance13:30
Concept Pairs, Google Distance13:38
Approaches14:07
Correlative Multi - Label Learning14:28
Tag Concurrence Distance14:28
Automated Annotation - 1st Paradigm14:38
Concept Pairs15:15
To Exploit Label Correlations15:49
Different Concept Relationships16:05
Concept Distance17:00
Automated Annotation - 2nd Paradigm - 117:08
Can we mine concept distance from image content?17:18
Some Facts17:26
Overview of Flickr Distance18:05
Automated Annotation - 2nd Paradigm - 218:29
Automated Annotation - 3rd Paradigm 18:39
CML Roadmap19:03
Flickr Distance - 119:11
Automated Annotation - 3rd Paradigm19:29
What We Need - 119:47
How to Model Concept Correlations19:59
What We Need - 220:46
Correlative Multi - Label Video Annotation - 120:55
What We Need - 321:36
Statistical Language Model21:48
Correlative Multi - Label Video Annotation - 221:57
Visual Language Model 22:36
Correlative Multi - Label Video Annotation - 323:42
Performance of VLM23:46
Latent - Topic VLM - 124:15
Correlative Multi - Label Video Annotation - 424:22
Latent - Topic VLM - 224:39
Correlative Multi - Label Video Annotation - 524:51
Performance of LT - VLM25:08
Flickr Distance - 225:40
Correlative Multi - Label Video Annotation - 625:59
Procedure of Flickr Distance26:25
Correlative Multi - Label Video Annotation - 726:54
Experiments26:56
Experiments - Configurations27:00
Correlative Multi - Label Video Annotation - 827:19
Eva1: Subjective Evaluation27:28
Correlative Multi - Label Video Annotation - 927:47
Correlative Multi - Label Video Annotation - 1028:00
Eva2: Objective Evaluations28:24
App1: Concept Clustering28:53
Correlative Multi - Label Video Annotation - 1129:07
App2: Image Annotation29:25
Correlative Multi - Label Video Annotation - 1229:43
App3: Tag Recommendation30:09
Outline Active Multi - Label Learning30:15
Learning Based Tagging30:32
Discussion30:34
Where Is the Way Out?31:07
Similar Patterns31:32
Flickr Distance - 332:13
Social Tagging and Tag Processing32:34
A Promisin Direction - Active Annotation33:08
An Attempt34:00
Test - 234:25
Online Multi - Label Active Learning - 135:18
Online Multi - Label Active Learning - 235:35
Online Multi - Label Active Learning - 335:48
Outline - 835:53
Outline - 935:57
Online Multi - Label Active Learning - 436:00
Online Multi - Label Active Learning - 536:33
Data Driven Web - Scale Tagging36:57
Online Multi - Label Active Learning - 637:30
Introduction37:34
Example - 438:22
Labels38:29
Multi - Label (Two - Dimensional) Active Learning38:56
Example - 539:27
How Many Duplicates on the Web?39:49
Online Multi - Label Learner - 139:51
Online Multi - Label Learner - 240:14
CBIR - Based Tagging - 140:17
CBIR - Based Tagging - 240:28
Experiments - 140:33
Near-dup search based Tagging on 2 Billion Images - Basic Idea41:00
Experiments - 241:06
Thr Value of Duplicates41:22
Experiments - 341:23
Active Annotations42:04
How Learning - Based Tagging is Used42:26
DEMOS42:48
Example - 642:54
Example - 743:20
Example - 143:33
Example - 843:35
Example - 943:49
Example - 1044:15
Some Numbers44:32
Learning - Based Tagging44:41
Towards Efficient Near-dup Search45:21
Test - 146:00
Hash Code Based Near-dup Detection47:01
Summary - Dup search methods47:35
Building Web - Scale Image Graph48:32
Techniques - Scalable k-NN graph construction51:16
Outline - 552:27
Outline - 652:40
Social Tagging and Tag Processing52:47
Social Media and the Associated Tags52:50
Key technologies53:13
Random divide-and-conquer53:29
Illustration of a random division53:37
Social tags are good, but they are:53:49
Multiple random divide-and-conquer54:46
Illustration of multiple divisions55:02
Tag Ranking55:10
The most relevant tag is not at the top position55:18
This phenomenon is widespread on social media websites such as Flickr55:54
Search on Flickr56:32
Illustration of neighborhood propagation - 157:06
What are we going to do:57:10
But how can we make it?57:40
Image Ranking vs. Tag Ranking57:45
Illustration of neighborhood propagation - 257:56
Illustration of neighborhood propagation - 358:12
Illustration of neighborhood propagation - 458:26
Illustration of neighborhood propagation - 558:26
Image Reranking vs. Tag Ranking58:28
Illustration of neighborhood propagation - final58:29
Can we borrow some idea from image reranking?59:18
Quantative evaluation59:18
Basic Assumptions59:26
Typical Image ReRanking - Random Walk59:50
Demo01:00:11
Basic Assumptions01:00:21
Tag Ranking (for each image)01:00:46
Example - 1101:00:55
Teh Problem is:01:00:57
What We Can Use - 101:01:14
WordNet Distance01:01:19
Example - 1201:01:58
Google Distance01:02:10
Real Demo01:02:11
What We Can Use - 201:02:57
Tag Concurrence Distance01:03:12
Example - 1301:03:55
Example - 1401:03:56
Example - 1501:03:58
Not Finished Yet01:03:59
Example - 201:04:09
Tag 2Image Distance01:04:21
Test - 301:04:28
Random Walk Based Tag Ranking - 101:06:35
Random Walk Based Tag Ranking - 201:06:46
Random Walk Based Tag Ranking - 301:06:56
A Better Measure: Flickr Distance01:07:30
Is It Enough?01:07:32
Basic Assumptions01:07:35
Initial Relevance Estimation01:08:31
Outline - 1001:09:13
Outline - 1101:09:20
Evolution of Extracting Semantics01:09:21
Idea Solution01:09:58
Rethinking Bridging Semantic Gap - 101:09:59
Initial Rank Estimation01:10:12
In Summary: Tag Ranking01:10:36
Performance Evaluation01:10:51
Rethinking Bridging Semantic Gap - 201:11:10
Example - 301:11:24
Rethinking Bridging Semantic Gap - 301:11:45
After Tag Ranking:01:12:14
A Preliminary Attempt01:12:15
Application 1: Tag - based search - 101:12:25
Application 1: Tag - based search - 201:12:44
Performance of Tag - Based Search01:12:49
Application 2: Auto Tagging01:13:05
Study of the China Market01:13:38
Performance of Auto Tagging01:13:56
Application 3: Group recomendation01:14:55
Rethinking Bridging Semantic Gap - 401:15:01
Performance of Group Recommendation 01:15:09
Conclusion01:15:30
Future Directions01:15:39
Outline - 701:16:03
Questions?01:16:15