Multiple Hypothesis Tracking Revisited

By Chanho Kim, Fuxin Li, A. Ciptadi, James M. Rehg. IEEE International Conference on Computer Vision, 2015

651 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/6968cc16bccd99c3d21d09b226234e639002c07e

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 Computer Vision and Pattern Recognition

Citations in the corpus

MOTS: Multi-Object Tracking and Segmentation (2019, result): “Despite these odds, our method achieves results close to jCC [21] and MHT-DAM [22].”
HOTA: A Higher Order Metric for Evaluating Multi-object Tracking (2020, methodology): “2 MHT_DAM_17 [34] 21 14 (↓ 7) 23 (↑ 2) 38.”
Online Multi-Object Tracking with Dual Matching Attention Networks (2018, neutral): “9 MHT DAM[23] 50.”
Simple online and realtime tracking (2016, methodology): “To aid the data association process, trackers use various methods for modelling the motion [1, 4] and appearance [5, 3] of objects in the scene.”
Segment as Points for Efficient Online Multi-Object Tracking and Segmentation (2020, neutral): “MHT-DAM [16]+ MG 48.”
MOTChallenge: A Benchmark for Single-Camera Multiple Target Tracking (2020, methodology): “MHT_bLSTM (Kim et al. 2018) replaces the multi-output regularized least-squares learning framework of MHT_DAM (Kim et al. 2015) with a bi-linear LSTM and adapts both the appearance model as well as the convolutional ﬁlters in an online fashion.”
Deep learning in video multi-object tracking: A survey (2019, methodology): “Kim et al. [93] proposed the use of a so-called Bilinear LSTM network as the gating step of the MHT-DAM [83] algorithm, that is, the affinity score computed by the LSTM was used to decide whether to prune or not a certain branch of the hypotheses tree.”
UA-DETRAC: A new benchmark and protocol for multi-object detection and tracking (2015, methodology): “Followed by the EB + KIOU method, the EB + IOU and Faster R-CNN + MHT methods achieve higher PR-MOTA scores ( i.e. , 19 .”
Quasi-Dense Similarity Learning for Multiple Object Tracking (2020, methodology): “Appearance similarity in MOT To exploit instance appearance similarity to strengthen tracking and re-identify vanished objects, some methods directly use an independent model [2,20,22,29,38,40,45,49] or add an extra embedding head to the detector for end-to-end training [26,44,50,53].”
Multiple People Tracking by Lifted Multicut and Person Re-identification (2017, background): “We compare our method with the best published results on the benchmark, including NOMT[6], MHT-DAM [14], OVBT [2],LTTSC-CRF [16], CEM [21], TBD [10] and Multicut [28].”
TrackFormer: Multi-Object Tracking with Transformers (2021, methodology): “For MOTS20 we average validation metrics over all splits and report the results from a single epoch (which yields the best mean MOTA / MOTSA) over all splits, i.e ., we do not take the best epoch for each individual split.”
Pose Flow: Efficient Online Pose Tracking (2018, background): “[ Kim et al. , 2015; Choi, 2015] or grouping detections into tracklets and then merge them into tracks [Bing Wang, 2015].”
Joint Graph Decomposition and Node Labeling by Local Search (2016, background): “This problem offers a common mathematical abstraction of seemingly unrelated computer vision tasks, including instance-separating semantic segmentation, articulated human body pose estimation and multiple object tracking.”
Joint Graph Decomposition & Node Labeling: Problem, Algorithms, Applications (2016, methodology): “Multiple object tracking [3, 4, 6, 9, 17, 22, 24, 35, 36] can be seen as a task requiring two classes of decisions: For every point in an image, one needs to decide whether this point depicts an object or background.”
Spatial-Temporal Relation Networks for Multi-Object Tracking (2019, neutral): “0.5 19,993 281,643 5,988 7,404 36.5 MHT DLSTM [26] 47.5 77.5 51.9 71.4 40.8 18.2 41.7 25,981 268,042 2,069 3,124 28.8 EDMT [8] 50.0 77.3 51.3 67 41.5 21.6 36.3 32,279 247,297 2,264 3,260 24.0 MHT DAM [25] 50.7 77.5 47.2 63.4 37.6 20.8 36.9 22,875 252,889 2,314 2,865 25.4 jCC [22] 51.2 75.9 54.572.24”
Towards Real-Time Multi-Object Tracking (2019, background): “1) Ones that model the association problem as certain form of optimization problem on graphs [37,42,17].”
Deep Affinity Network for Multiple Object Tracking (2018, methodology): “The approaches following this pipeline exploit several distinct types of representation models, including; appearance models [10], [11], [12], [13], motion models [14], [15], [16], [17], and composite models [9], [18], [19].”