schema participation in trecvid 2004schema ist-2001-32795 28 conclusions •performance of schema-xm...
TRANSCRIPT
![Page 1: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/1.jpg)
SCHEMA Participationin TRECVID 2004
Vasileios Mezaris1, Haralambos Doulaverakis1, Stephan Herrmann2, Bart Lehane3, Noel O’Connor3, Ioannis Kompatsiaris1, and Michael G. Strintzis1.
1Informatics and Telematics Institute (ITI)/Centre for Researchand Technology Hellas (CERTH)
2Institute for Integrated Systems, Munich University of Technology
3Centre for Digital Video Processing, Dublin City University, Ireland
![Page 2: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/2.jpg)
SCHEMA IST-2001-32795 2
Outline
• Background• Canned system demo• Under the hood:
Image analysis MPEG-7 XM Face classifier Motion features
• Experimental setup & results• Conclusions
![Page 3: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/3.jpg)
SCHEMA IST-2001-32795 3
SCHEMA NoE• EU-funded Network of Excellence
Network of Excellence in Content-Based SemanticScene Analysis and Information Retrieval
• 13 members & 28 affiliated members Universities, research institutes, companies (SMEs
& MNCs)
• Activities: Researcher mobility (collaboration, exchanges) Contributions to standards (MPEG-7, JPEG200) Development of a Reference System for CBIR
![Page 4: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/4.jpg)
SCHEMA IST-2001-32795 4
SCHEMA participation inTRECVID
• Adapt Reference System for TRECVIDexperiments
• 2 Variants: Text-based searching on ASR only Text & image-based features:
Option 1: initiate search with text & high-level features… use visual similarity thereafter
Option 2: initiate search with text, high-level featuresand example image … use visual similarity thereafter
![Page 5: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/5.jpg)
SCHEMA IST-2001-32795 5
![Page 6: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/6.jpg)
SCHEMA IST-2001-32795 6
![Page 7: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/7.jpg)
SCHEMA IST-2001-32795 7
![Page 8: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/8.jpg)
SCHEMA IST-2001-32795 8
![Page 9: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/9.jpg)
SCHEMA IST-2001-32795 9
![Page 10: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/10.jpg)
SCHEMA IST-2001-32795 10
![Page 11: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/11.jpg)
SCHEMA IST-2001-32795 11
![Page 12: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/12.jpg)
SCHEMA IST-2001-32795 12
![Page 13: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/13.jpg)
SCHEMA IST-2001-32795 13
![Page 14: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/14.jpg)
SCHEMA IST-2001-32795 14
![Page 15: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/15.jpg)
SCHEMA IST-2001-32795 15
![Page 16: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/16.jpg)
SCHEMA IST-2001-32795 16
Acquisition of results• In each variant and option, when the
user found a keyframe/shot that wasrelevant to the topic, he would add it tothe result set by clicking on it Process similar to a shopping cart
application
![Page 17: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/17.jpg)
SCHEMA IST-2001-32795 17
Image Analysis
• 2 region-basedsegmentation algorithms: Modified RSST (MRSST)
Bottom-up colour-based pixelmerging
K-Means-with-Connectivity-Constraint (KMCC)
Hierarchical k-means clusteringof colour & texture features
MRSST
KMCC
![Page 18: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/18.jpg)
SCHEMA IST-2001-32795 18
MPEG-7 XM
• MPEG-7 eXperimentation Model (XM)used for: Low-level feature extraction for images or
regions Descriptor matching for search & retrieval
• XM requires 3 modifications to make itsuitable for use in a “real” IR system
![Page 19: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/19.jpg)
SCHEMA IST-2001-32795 19
XM Modification 1• MultiImage module:
New module to support extraction and matchingof multiple descriptors
Normalisation of working point of matchingfunctions
Linear combination of distance values usingdefault weights
![Page 20: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/20.jpg)
SCHEMA IST-2001-32795 20
XM Modification 2
• XM Server Re-engineer software to allow XM to run
continuously in the background, accessed via TCPsocket communication
![Page 21: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/21.jpg)
SCHEMA IST-2001-32795 21
XM Modification 3
• Indexing Use pre-computed similarity values Hierarchically partition search space
![Page 22: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/22.jpg)
SCHEMA IST-2001-32795 22
Face Classifier• J. Luo and A.
Savakis [ICIP 01]• Extension:
Apply imagesegmentation
Classifier trained onsegmented faceregions
Significantimprovement fordistant faces
![Page 23: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/23.jpg)
SCHEMA IST-2001-32795 23
Motion Features
• MPEG-7 Motion intensity: “Very Low”, “Low”, “Medium”, “High”,
“Very High”
• Global camera motion: Runs of 0 motion vectors in MPEG-1 P-
frames “Low”, “Medium”, “High” estimated using a
Fuzzy C-Means algorithm
![Page 24: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/24.jpg)
SCHEMA IST-2001-32795 24
Experiments
• 8 users, each searching for 12 topics 6 using text-only, 6 using text & image
features• 10 minutes were given to each tester to
find as many shots as possible, relevantto the given topic
• Conducted remotely in 3 different labs• 2 runs for each system submitted
![Page 25: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/25.jpg)
SCHEMA IST-2001-32795 25
Evaluation
![Page 26: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/26.jpg)
SCHEMA IST-2001-32795 26
Evaluation
![Page 27: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/27.jpg)
SCHEMA IST-2001-32795 27
EvaluationSCHEMA vs Median (run XM_1)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
125
127
129
131
133
135
137
139
141
143
145
147
Topic
Av.
Pr
ecis
ion
MedianXM_1
SCHEMA vs Median (run XM_2)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
125
127
129
131
133
135
137
139
141
143
145
147
Topic
Av.
Pr
ecis
ion
MedianXM_2
SCHEMA vs Median (run Text_1)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
125
127
129
131
133
135
137
139
141
143
145
147
Topic
Av.
Pr
ecis
ion
MedianText_1
SCHEMA vs Median (run Text_2)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
125
127
129
131
133
135
137
139
141
143
145
147
Topic
Av.
Pr
ecis
ion
MedianText_2
![Page 28: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/28.jpg)
SCHEMA IST-2001-32795 28
Conclusions
• Performance of SCHEMA-XM better thanSCHEMA-Text Inclusion of low-level features, regions and
high-level “filters” aids retrieval for certaintopics
• Performed reasonably well incomparison to median … for first timers
![Page 29: SCHEMA Participation in TRECVID 2004SCHEMA IST-2001-32795 28 Conclusions •Performance of SCHEMA-XM better than SCHEMA-Text Inclusion of low-level features, regions and high-level](https://reader034.vdocument.in/reader034/viewer/2022042913/5f49b8c7b9362426c76b6b86/html5/thumbnails/29.jpg)
SCHEMA IST-2001-32795 29
Thank you!
• SCHEMA Reference System: http://media.iti.gr/SchemaRS/
SCHEMA Ref Sys (Corel DB) SCHEMA XM (Corel DB) SCHEMA MPA (Macedonian Press Agency DB)