… We present a generic event detection system for the SED task of TRECVID 2014. It consists
of two parts: the retrospective system and the interactive system. The retrospective system …

… We present a generic event detection system for the SED task of TRECVID 2014. It consists
of two parts: the retrospective system and the interactive system. The retrospective system …

… In this section, we study our TRECVID 2013 MED system and find that: ➢ Semantic features
trained by deep neural network can outperform the best low-level features we currently have …

… In this abstract paper we summarize Informedia’s system for TRECVid 2018 Ad-hoc Video
Search (AVS). We developed systems using two types of representations for retrieval followed …

… TRECVID is aimed to retrieve videos that contains the described events; the surveillance event
detection (SED) task at TRECVID … VH at MediaEval and TRECVID show that the unimodal …

We propose an action recognition system for surveillance scenarios, which wins TRECVID
2020 [1] Activities in Extended Video (ActEV2) Challenge with a large advantage of 23.8% …

In TRECVID 2021 ActEV task, we tackle multi-scale multi-instance activity detection in
extended videos with the CMU Argus++[28] video understanding framework. The proposed …

… In this paper, we provide an overview of the BBN’s VISER system for the TRECVID 2014
MED and MER evaluations. Our system uses a combination of low-level visual, audio and …

Waseda participated in the TRECVID 2015 Semantic Indexing (SIN) task [6]. For the SIN task,
our approach used the following processing pipelines: feature extraction using several …

… The main difference between TRECVID MED 2014 and 2015 is that we were allowed to
provide five runs in the Pre-Specified (PS) case and two runs in the Ad-Hoc (AH) case. With …