Graduate School and Research Center in Digital Sciences

Uploader models for video concept detection

Merialdo, Bernard; Niaz, Usman

CBMI 2014, 12th International Workshop on Content-Based Multimedia Indexing, 18-20 June 2014, Klagenfurt, Austria

In video indexing, it has been noticed that a simple uploader model was able to improve the MAP of concept detection in the TRECVID Semantic Concept Indexing (SIN) task. In this paper, we explore this idea further by comparing different types of uploader models and different types of score/rank distribution. We evaluate the performance of these combinations on the best SIN 2012 runs, and explore the impact of their parameters. We observe that the improvement is generally lower for the best runs than for the weaker runs. We also observe that tuning the models for each concept independently produces a much more significant improvement.

Document Doi Bibtex

Title:Uploader models for video concept detection
Keywords:Multimedia Indexing, User model, TRECVID
Department:Data Science
Eurecom ref:4294
Copyright: © 2014 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Bibtex: @inproceedings{EURECOM+4294, doi = {}, year = {2014}, title = {{U}ploader models for video concept detection}, author = {{M}erialdo, {B}ernard and {N}iaz, {U}sman }, booktitle = {{CBMI} 2014, 12th {I}nternational {W}orkshop on {C}ontent-{B}ased {M}ultimedia {I}ndexing, 18-20 {J}une 2014, {K}lagenfurt, {A}ustria}, address = {{K}lagenfurt, {AUSTRIA}}, month = {06}, url = {} }
See also: