Abstract
We participated in one of the two types of Instance Search task in TRECVID 2019: Fully Automatic Search, without any human intervention. Firstly, the specific person and action are searched separately, and then we re-rank the two sorts of search results by ranking the one type scores according to the other type, as well as the score fusion. And thus, three kinds of final instance search results are submitted. Specifically, for the person search, our baseline consists of face detection, alignment and face feature selection. And for the action search, we integrate person detection, person tracking and feature selection into a framework to get the final 3D features for all tracklets in video shots. The official evaluations showed that our best search result gets the 4th place in the Automatic search.
Original language | English |
---|---|
Title of host publication | TREDVID 2019 |
Editors | George Awad |
Place of Publication | Saarbrücken/Wadern Germany |
Publisher | Schloss Dagstuhl |
Number of pages | 4 |
Publication status | Published - 2019 |
Event | TREC Video Retrieval Evaluation 2019 - Gaithersburg, United States of America Duration: 12 Nov 2019 → 13 Nov 2019 https://dblp.org/db/conf/trecvid/trecvid2019.html (Proceedings) https://www-nlpir.nist.gov/projects/tv2019/schedule.html (Website) |
Conference
Conference | TREC Video Retrieval Evaluation 2019 |
---|---|
Abbreviated title | TRECVID 2019 |
Country/Territory | United States of America |
City | Gaithersburg |
Period | 12/11/19 → 13/11/19 |
Internet address |