Version 16.0
|
 |
Developed from August, 2017 till January, 2018 by ITI. VERGE v16.0 participated in Video Browser Showdown.
Functionalities provided:
- High Level Concept Retrieval
- Visual Similarity Search
- Text-based Search
- Video Clustering
- Automatic Query Formulation & Expansion
- Multimodal Video Fusion
- A dashboard menu
- Shot-based or video-based representation of results
- A film-strip that shows the complete video of a shot
Visit Verge Engine v16.0.
|
|
Version 14.0
|
 |
Developed in first semester of 2016 by ITI. VERGE v14.0 participated in Video Browser Showdown
and took the 4th place out of 9 participants.
Functionalities provided:
- Color-based search
- Visual Similarity search
- Visual concept-based search
- Search by image example
- Endless scrolling
- Fast video navigation
Visit Verge Engine v14.0 for a demonstration
& Github link for the source code.
|
|
Version 13.0
|
 |
Developed in first semester of 2015 by ITI. VERGE v13.0 participated in TRECVID 2015
in the Instance search (INS) & Surveillance event detection (SED) tasks.
Functionalities provided:
- Visual Similarity search
- Visual concept-based search
- Object-based retrieval
- Search by image example
- Endless scrolling
- Fast video navigation
Visit Verge Engine v13.0
|
|
Version 12.0
|
 |
Developed in September 2012 by ITI. VERGE v12.0 participated in TRECVID 2012.
Involved dataset is TRECVID 2012 Known Item Search test data.
Functionalities provided:
- Text search on ASR and metadata data
- Visual Similarity by Example based on MPEG-7 descriptors
- High Level Visual Concepts and ontology-tree representation
- Video Similarity by Example based on Textual Metadata and PLSA
- Video and shot based presentation
- Keyframes presentation for videos
- Basket for selected shots
Visit Verge Engine v12.0
|
|
Version 11.0
|
 |
Developed in September 2011 by ITI. VERGE v11.0 participated in TRECVID 2011.
Involved dataset is TRECVID 2011 Known Item Search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- High Level Visual Concepts and ontology-tree representation
- Video Similarity by Example based on Textual Metadata and PLSA
- Fusion of High Level Visual and textual Concepts based on implicit user feedback
- Video shots presentation
- Basket for selected shots
Visit Verge Engine v11.0
|
|
Version 10.1
|
 |
Developed in August 2010 by ITI. VERGE v9.1 participated in TRECVID 2010.
Involved dataset is TRECVID 2010 Known Item Search (KIS) test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Manually assisted Fusion of High Level Visual and Textual Concepts
- High Level Visual Concepts and ontology-tree representation
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Color filtering
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
Visit Verge Engine v10.1
|
|
Version 10.0
|
 |
Developed in August 2010 by ITI. VERGE v10.1 participated in TRECVID 2010.
Involved dataset is TRECVID 2009 Instanse Search (INS) test data.
Functionalities provided:
- Automatic Image segmentation and visual segment search
- Visual Similarity by Example based on MPEG-7 descriptors
- High Level Visual Concepts and ontology-tree representation
- Manually assisted Fusion of High Level Visual and Textual Concepts
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Color filtering
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
|
|
Version 9.1
|
 |
Developed in August 2009 by ITI. VERGE v9.1 participated in TRECVID 2009.
Involved dataset is TRECVID 2009 search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Visual Similarity by Example based on SIFT descriptor
- Manually assited fusion of High Level Visual and Textual Concepts
- High Level Visual Concepts and ontology-tree representation
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Color filtering
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
|
|
Version 9.0
|
 |
Developed in May 2009 by ITI. VERGE v9.0 participated in VideOlympics 2009.
Involved dataset is TRECVID 2008 search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Visual Similarity by Example based on SIFT descriptor
- High Level Visual Concepts combined with ontology
- High Level Text Concepts
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Color filtering
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
|
|
Version 8.2
|
 |
Developed in August 2008 by ITI with the support of COST action. Specific modules were implemented by QMUL and University of Zilina. VERGE v8.2 participated in TRECVID 2008.
Involved dataset is TRECVID 2008 search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Visual Similarity by Example based on SIFT descriptor
- High Level Visual Concepts
- High Level Text Concepts
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Audio filtering (University of Zilina)
- Relevance Feedback based on visual features (QMUL)
- Color filtering
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
|
|
Version 8.1
|
 |
Developed in May 2008 by ITI. VERGE v8.1 participated in VideOlympics 2008.
Involved dataset is TRECVID 2007 search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Visual Similarity by Example based on SIFT descriptor
- High Level Visual Concepts
- High Level Text Concepts
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
|
|
Version 8.0
|
 |
Intermediate version developed in May 2008 by ITI for experimental purposes.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Visual Similarity by Example based on SIFT descriptor
- High Level Visual Concepts
- High Level Text Concepts
- Related keywords based on Wordnet synsets, hypernyms and hyponyms
- Sideshots presentation
- Video shots presentation
- Basket for selected shots
|
|
Version 7.1
|
 |
Developed in August 2007 by ITI with the support of COST action. Specific modules were implemented by QMUL and University of Belgrade. VERGE v7.1 participated in TRECVID 2007.
Involved dataset is TRECVID 2007 search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Simple text search
- Related keywords based on words coexistence
- Color filtering
- Relevance Feedback based on visual features (QMUL, University of Belgrade)
- Sideshots presentation
- Basket for selected shots
|
|
Version 7.0
|
 |
Developed in May 2007 by ITI. VERGE v7.0 participated in VideOlympics 2007.
Involved dataset is TRECVID 2006 search test data.
Functionalities provided:
- Visual Similarity by Example based on MPEG-7 descriptors
- Simple text search
- Related keywords based on words coexistence
- Sideshots presentation
- Basket for selected shots
|
|
