People-tracking-by-detection and people-detection-by-tracking

By Mykhaylo Andriluka, Stefan Roth, Bernt Schiele. 2008

1,005 citations recorded by the source corpus.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Biomedical Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Video In Sentences Out (2012, background): “While such information is far noisier and less accurate than fitting precise articulated models (Andriluka et al., 2008; Bregler, 1997; Gavrila and Davis, 1995; Sigal et al., 2010; Yang and Ramanan, 2011) and appears unintelligible to the human eye, as shown in Section 3.”
Pedestrian detection: A benchmark (2009, neutral): “MIT[27] 924 – – – – – 128 128 128 3 USC-A[43] – – – 313 – 205 70 98 133 3 USC-B[43] – – – 271 – 54 63 90 126 3 USC-C[44] – – – 232 – 100 74 108 145 3 3 CVC[14] 1000 6175† – – – – 46 83 164 3 3 TUD-det[1] 400 – 400 311 – 250 133 218 278 3 3 INRIA[5] 1208 1218 614 566 453 288 139 279 456 3 DC[24] 2.”
Vision Meets Drones: A Challenge (2018, neutral): “9k 640× 480 2007 TUD [15] life 1, 818 1 3, 274 640× 480 2008 EPFL Multi-View Car [16] exhibition 2, 000 1 2, 000 376× 250 2009 Caltech Pedestrian [1] driving 249k 1 347k 640× 480 √ 2012 KITTI Detection [2] driving 15.”
Towards 3D object recognition via classification of arbitrary object tracks (2011, background): “In general, object recognition in dynamic scenes using classspecific trackers will (presumably) obtain better performance for the classes they are designed for, but at the cost of additional expert time to specify new models if they are needed.”
Object Detection from Video Tubelets with Convolutional Neural Networks (2016, background): “The large number of tracking-by-detection methods [1, 2, 3, 26] for multi-pedestrian tracking have shown that temporal information could be utilized to regularize the detection re-sults.”
Discrete-continuous optimization for multi-target tracking (2012, methodology): “Three sequences (Campus, Crossing and Stadtmitte) are taken from the TUD dataset [1].”
Alternating Regression Forests for Object Detection and Pose Estimation (2013, methodology): “Experimental Setup: We use three object detection data sets commonly employed for evaluating HFs, namely the TUD-pedestrian, TUD-crossing and TUD-campus benchmarks [2].”
Argoverse: 3D Tracking and Forecasting With Rich Maps (2019, background): “Many existing datasets for object tracking focus on pedestrian tracking from image/video sequences [32, 28, 2].”
SAIL-VOS: Semantic Amodal Instance Level Video Object Segmentation – A Synthetic Dataset and Baselines (2019, methodology): “Reference images are used in [39] and the approach is evaluated on the TUD Pedestrian [3] and the Polo dataset [107].”
Multi-target tracking by on-line learned discriminative appearance models (2010, methodology): “There are two main streams in detection-based tracking methods: one considers only past and current frames to make association decisions [7, 8, 20, 25, 27]; the other takes information from future frames also [ 3 , 5, 15, 16, 18, 21, 26, 28].”
We Are Family: Joint Pose Estimation of Multiple Persons (2010, methodology): “As in several recent HPE methods [1,12,18], we use a generic person detector to determine the approximate location and scale of the persons in an image.”
Multiobject tracking as maximum weight independent set (2011, methodology): “Errors occur when a target person is: (i) very close to other targets in the ETH Central, TUD Crossing sequences; (ii) sitting in the ETH Central videos; or (iii) partially out of the field of view in the i-Lids videos.”
UA-DETRAC: A new benchmark and protocol for multi-object detection and tracking (2015, neutral): “9k 742 182k ✓ ✓ T P, C ✓ ✓ 2016 TUD (Andriluka et al., 2008) 610 – 610 451 31 2.”
Hough Forests for Object Detection, Tracking, and Action Recognition (2011, methodology): “While [65], [67] are based on an explicit model with joint annotation, the ISM-based approach [68] relies on silhouettes.”