This paper describes the submissions of the EURECOM team to the TrecVid 2018 VTT task. We participated in the Sentence Matching subtask. Our approach is to project both descriptive texts and videos in the same vector space through a deep neural network, and to compare them using a cosine similarity. In particular, we compare several variants of sentence embeddings.
EURECOM participation in TrecVid VTT 2018
TRECVID 2018, 22nd International Workshop on Video Retrieval Evaluation, November 13-15, 2018, Gaithersburg, USA
© NIST. Personal use of this material is permitted. The definitive version of this paper was published in TRECVID 2018, 22nd International Workshop on Video Retrieval Evaluation, November 13-15, 2018, Gaithersburg, USA and is available at :
PERMALINK : https://www.eurecom.fr/publication/5753