Ecole d'ingénieur et centre de recherche en Sciences du numérique

IRIM at TRECVID 2015: Semantic indexing

Le Borgne, Hervé; et al.

TRECVID 2015, 19th International Workshop on Video Retrieval Evaluation, 16-18 November 2015, Gaithersburg, MD, USA

The IRIM group is a consortium of French teams supported by the GDR ISIS and working on Multimedia Indexing and Retrieval. This paper describes its participation to the TRECVID 2015 semantic indexing (SIN). 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, classi cation, fusion of descriptor variants, higher-level fusion, and re-ranking. We evaluated a number of di erent de- scriptors and tried di erent fusion strategies. The best IRIM run has a Mean Inferred Average Precision of 0.2947, which ranked it 4th out of 15 participants.

Document Hal Bibtex

Titre:IRIM at TRECVID 2015: Semantic indexing
Type:Conférence
Langue:English
Ville:Gaithersburg
Pays:ÉTATS-UNIS
Date:
Département:Data Science
Eurecom ref:4731
Copyright: © NIST. Personal use of this material is permitted. The definitive version of this paper was published in TRECVID 2015, 19th International Workshop on Video Retrieval Evaluation, 16-18 November 2015, Gaithersburg, MD, USA and is available at :
Bibtex: @inproceedings{EURECOM+4731, year = {2015}, title = {{IRIM} at {TRECVID} 2015: {S}emantic indexing}, author = {{L}e {B}orgne, {H}erv{\'e} and et al.}, booktitle = {{TRECVID} 2015, 19th {I}nternational {W}orkshop on {V}ideo {R}etrieval {E}valuation, 16-18 {N}ovember 2015, {G}aithersburg, {MD}, {USA}}, address = {{G}aithersburg, {\'{E}}{TATS}-{UNIS}}, month = {11}, url = {http://www.eurecom.fr/publication/4731} }
Voir aussi: