Evaluation campaigns and TRECVid

By Alan F. Smeaton, Paul Over, Wessel Kraaij. Multimedia Information Retrieval, 2006

1,428 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/3cb1f624b56eca597d65d27b8da7c4f2cd4b8531

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

LabelMe: Online Image Annotation and Applications (2010, background): “containing many hours of television or surveillance data [63], [2], [3], [17], [59].”
Learning to associate: HybridBoosted multi-target tracker for crowded scene (2009, methodology): “We applied our approach to human tracking and carried out the experiments on the CAVIAR dataset [13] and the TRECVID08 dataset [14] which features a much more challenging airport scene.”
Computing (2021, methodology): “In this chapter, we are highlighting outcomes from these efforts, for full details of the training procedures, please see Yates [2018].”
Finding meaning on YouTube: Tag recommendation and category discovery (2010, methodology): “It is important to note that our method differs from the traditional learning problems on video, such as those described in the TRECVID [25] in the sense that we do not have a “constrained” number of labels.”
Unified Video Annotation via Multigraph Learning (2009, methodology): “The National Institute of Standards and Technology (NIST) has also established " highlevel feature extraction " as a task in TREC video retrieval evaluation (TRECVID) [1], [28], which aims to provide a benchmark for evaluating video annotation technologies.”
Attribute-Based Classification for Zero-Shot Visual Object Categorization (background): “For example, the TRECVID contest [33] contains a task of high-level feature extraction, which consists of predicting semantic concepts, in particular scene types, e.g. outdoor, urban, and high-level actions, e.g. sports.”
Sparselet Models for Efficient Multiclass Object Detection (2012, methodology): “We manually selected a set of object categories which are related to events in the 2011 TRECVID MED challenge [22]: sailboat, bread, cake, candle, fish, goat, jeep, scissors, and tire.”
Aligning plot synopses to videos for story-based retrieval (2015, methodology): “Many popular tasks have been proposed under the TRECVid [38] evaluation campaign.”
People reidentification in surveillance and forensics (2013, background): “Datasets Available for People Reidentification Name & Ref Image/Video People Additional info ViPER Still Images 632 Scenario: Outdoor Place: Outdoor surveillance [Gray et al. 2007] People Size: 128×48 vision.soe.ucsc.edu I-LIDS Video [fps=25] Scenario: Outdoor/Indoor 5 cameras PAL 1000 Place: Collec”
Efficient Classification for Additive Kernel SVMs (2013, methodology): “Many state-of-the-art object detection and image classification results on the PASCAL Visual Object Challenge [12], ImageNet [28], and TRECVID [53] challenge are based on variants of the kernel where the underlying features change.”
Detection and Classification of Acoustic Scenes and Events (2015, background): “One further related challenge in audiovisual research is TRECVID Multimedia Event Detection, where the focus is on audiovisual, multi-modal event detection in video recordings [35].”
Real-time keyframe extraction towards video content identification (2009, methodology): “To evaluate the performance of the proposed keyframe extraction algorithm, eight MPEG-1 video sequences from TRECVid 2007 test video collection [12] where selected, namely BG 2402 (00:26:39)1, BG 26788 (00:14:21), BG 35046 (00:26:01), BG 35841 (01:01:43), BG 36878 (00:25:06), BG 37322 (00:26:11), BG”
Cross-domain learning methods for high-level visual concept classification (2008, background): “The TRECVID 2007 data set is quite different from TRECVID 2005 data set in program structure and production value, but they have similar semantic concepts of interest.”
The New Data and New Challenges in Multimedia Research (2015, methodology): “Using an online annotation tool people have been invited to draw the outlines of objects visible in the images, so far producing over 325,000 annotations for 5,650 object categories.”
Content-based visual search learned from social media (2012, methodology): “We adopt average precision (AP), a common choice for evaluating visual search engines [105].”
Learning to detect unseen object classes by between-class attribute transfer (2009, background): “In the field of multimedia retrieval, the annual TRECVID contest [32] contains a subtask of high-level feature extraction.”