IRIM - Indexation et Recherche d'Information Multimedia GDR-ISIS
Nicolas Ballas
,
Benjamin Labbé
,
Aymen Shabou
,
Hervé Le Borgne
,
Philippe Gosselin
,
Miriam Redi
,
Bernard Merialdo
,
Hervé Jégou
,
Jonathan Delhumeau
,
Rémi Vieux
,
Boris Mansencal
,
Jenny Benois-Pineau
,
Stéphane Ayache
,
Abdelkader Haadi
,
Bahjat Safadi
,
Franck Thollard
,
Nadia Derbas
,
Georges Quénot
,
Hervé Bredin
,
Matthieu Cord
,
Boyang Gao
,
Chao Zhu
,
Yuxing Tang
,
Emmanuel Dellandréa
(1)
,
Charles-Edmond Bichot
,
Liming Chen
(1)
,
Alexandre Benoit
,
Patrick Lambert
,
Tiberius Strat
,
Joseph Razik
,
Sébastion Paris
,
Hervé Glotin
,
Ngoc-Trung Tran
(2, 3)
,
Dijana Petrovska-Delacrétaz
,
Gérard Chollet
(4, 3)
,
Andrei Stoian
,
Michel Crucianu
Nicolas Ballas
- Function : Author
- PersonId : 767569
- IdRef : 176879218
Benjamin Labbé
- Function : Author
Aymen Shabou
- Function : Author
Hervé Le Borgne
- Function : Author
- PersonId : 181478
- IdHAL : herve-le-borgne
- ORCID : 0000-0003-0520-8436
- IdRef : 079208452
Philippe Gosselin
- Function : Author
Miriam Redi
- Function : Author
- PersonId : 765821
- IdRef : 171913272
Bernard Merialdo
- Function : Author
Hervé Jégou
- Function : Author
Jonathan Delhumeau
- Function : Author
Rémi Vieux
- Function : Author
Boris Mansencal
- Function : Author
Jenny Benois-Pineau
- Function : Author
- PersonId : 7842
- IdHAL : jenny-benois-pineau
- ORCID : 0000-0003-0659-8894
- IdRef : 074466992
Stéphane Ayache
- Function : Author
- PersonId : 16733
- IdHAL : stephane-ayache
- ORCID : 0000-0003-2982-7127
- IdRef : 129313254
Abdelkader Haadi
- Function : Author
Bahjat Safadi
- Function : Author
Franck Thollard
- Function : Author
Nadia Derbas
- Function : Author
Georges Quénot
- Function : Author
- PersonId : 3114
- IdHAL : georges-quenot
- ORCID : 0000-0003-2117-247X
- IdRef : 034104518
Hervé Bredin
- Function : Author
- PersonId : 15856
- IdHAL : hbredin
- ORCID : 0000-0002-3739-925X
- IdRef : 121165779
Matthieu Cord
- Function : Author
Boyang Gao
- Function : Author
Chao Zhu
- Function : Author
Yuxing Tang
- Function : Author
Emmanuel Dellandréa
- Function : Author
- PersonId : 7701
- IdHAL : emmanuel-dellandrea
- ORCID : 0000-0001-7346-228X
- IdRef : 114133034
Charles-Edmond Bichot
- Function : Author
- PersonId : 744760
- IdHAL : cebichot
Liming Chen
- Function : Author
- PersonId : 7562
- IdHAL : liming-chen
- IdRef : 067400175
Alexandre Benoit
- Function : Author
- PersonId : 6090
- IdHAL : alexandre-benoit
- ORCID : 0000-0002-0627-4948
- IdRef : 118249711
Patrick Lambert
- Function : Author
Tiberius Strat
- Function : Author
Joseph Razik
- Function : Author
- PersonId : 764010
- IdRef : 120118610
Sébastion Paris
- Function : Author
Hervé Glotin
- Function : Author
- PersonId : 181213
- IdHAL : herve-glotin
- IdRef : 112112501
Dijana Petrovska-Delacrétaz
- Function : Author
Gérard Chollet
- Function : Author
- PersonId : 176991
- IdHAL : gerard-chollet
- ORCID : 0000-0003-4245-146X
- IdRef : 078020824
Andrei Stoian
- Function : Author
Michel Crucianu
- Function : Author
Abstract
The IRIM group is a consortium of French teams working on Multimedia Indexing and Retrieval. This paper describes its participation to the TRECVID 2012 semantic indexing and instance search tasks. For the semantic indexing task, our approach uses a six-stages
processing pipelines for computing scores for the likelihood of a video shot to contain a target concept. These scores are then used for producing a ranked list of images or shots that are the most likely to contain the target concept. The pipeline is composed of the following steps: descriptor extraction, descriptor optimization, classification, fusion of descriptor variants, higher-level fusion, and re-ranking. We evaluated a number of different descriptors and tried dierent fusion strategies. The best IRIM run has a Mean Inferred Average Precision of 0.2378, which ranked us 4th out of 16 participants.
For the instance search task, our approach uses two steps. First individual methods of participants are used to compute similrity between an example image of instance and keyframes of a video clip. Then a two-step fusion method is used to combine these individual results and obtain a score for the likelihood of an instance to appear in a video clip. These scores are used to obtain a ranked list of clips the most likely to contain the queried instance. The best IRIM run has a MAP of 0.1192, which ranked us 29th on 79 fully automatic runs.