Stable multi-target tracking in real-time surveillance video

By Ben Benfold, Ian David Reid. Computer Vision and Pattern Recognition, 2011

725 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biomedical Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/5b1e33f60514a307054de5642a13051c1e1438b6

How researchers see this paper

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

In Artificial Intelligence

In Biomedical Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning scene-specific pedestrian detectors without real data (2015, background): “Town Center [3] (top), PETS 2006 [36] (middle) and CMUSRD [17] (bottom).”
SS-LSTM: A Hierarchical LSTM Model for Pedestrian Trajectory Prediction (2018, background): “Town Centre Dataset: The Town Centre dataset [2] contains hundreds of human trajectories in a real world crowded scene.”
Summarizing surveillance videos with local-patch-learning-based abnormality detection, blob sequence optimization, and type-based synopsis (2015, methodology): “Future work will include: (1) combining with trajectory association methods [12] to further improve the blob sequence extraction accuracy; (2) combining with automatic ROI detection methods [29,30] to automatically decide ROIs in a scene; (3) extending the activity detection module from abnormal act”
Unsupervised Object Discovery and Tracking in Video Collections (2015, background): “Conventional object tracking methods [41] usually require annotations for at least one frame [13, 15, 40], or object detectors trained for target classes in a supervised manner [1, 2, 26].”
PIE: A Large-Scale Dataset and Models for Pedestrian Intention Estimation and Trajectory Prediction (2019, background): “A number of datasets for trajectory prediction contain videos collected from a top-down view [18, 25, 22, 31] or surveillance camera perspective [23, 4, 45].”
Visual Tracking: an Experimental Survey (2013, neutral): “This category comprise of three videos from [1] with a fast-moving motorbike in the desert, a low contrast recording of a car on the highway, and a car-chase; three videos from the 3DPeS dataset [28] with varying illumination conditions and with complex object motion; one video from the dataset in [”
GMCP-Tracker: Global Multi-object Tracking Using Generalized Minimum Clique Graphs (2012, methodology): “We evaluated our method on four publicly available sequences, which provide a wide range of significant challenges: TUD-Crossing [16], TUD-Stadtmitte [17], Town Center [1], and sequence S2L1 from VS-PET2009 benchmark [3].”
A portable three-dimensional LIDAR-based system for long-term and wide-area people behavior measurement (2019, background): “There are several data sets which provide people behavior in indoor(6) and outdoor environments.(7,8) However, to the best of our knowledge, no data set provides people behavior involving interaction between followed and following persons even though such a situation is very common in daily services”
Forecasting Interactive Dynamics of Pedestrians with Fictitious Play (2016, methodology): “We collected ≈16K pedestrian patches from the Town Centre Dataset [5], with three labels for each patch, i.”
MOTChallenge 2015: Towards a Benchmark for Multi-Target Tracking (2015, methodology): “For the 4 sequences ﬁlmed using a static camera, AVG-TownCentre, PETS09-S2L1, PETS09-S2L2 and TUD-Stadtmitte, the calibration ﬁles from the sources [5], [10], [20] are used to compute a 2D homography between the image plane and the ground plane.”
Human motion trajectory prediction: a survey (2019, background): “Town Center Dataset (Benfold and Reid, 2011) Outdoor People Camera Pedestrians moving along a moderately crowded street 5 minutes, 230 hand labelled tracks Bounding boxes @15 Hz”
Computer vision and deep learning techniques for pedestrian detection and tracking: A survey (2018, background): “Recent literature contains several works dealing with marker-based human tracking, the most of which make use of drones [51,65,77–79, 91, 94, 96,”
On pairwise costs for network flow multi-object tracking (2014, methodology): “(a) & (b) Show performance for the MarchingRally and TownCenter sequences respectively.”