Robust Online Multi-object Tracking Based on Tracklet Confidence and Online Discriminative Appearance Learning

By Seung-Hwan Bae, Kuk-jin Yoon. 2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014

452 citations recorded by the source corpus.

Online multi-object tracking aims at producing complete tracks of multiple objects using the information accumulated up to the present moment. It still remains a difficult problem in complex scenes, because of frequent occlusion by clutter or other objects, similar appearances of different objects, and other factors. In this paper, we propose a robust online multi-object tracking method that can handle these difficulties effectively. We first propose the tracklet confidence using the detectability and continuity of a tracklet, and formulate a multi-object tracking problem based on the tracklet confidence. The multi-object tracking problem is then solved by associating tracklets in different ways according to their confidence values. Based on this strategy, tracklets sequentially grow with online-provided detections, and fragmented tracklets are linked up with others without any iterative and expensive associations. Here, for reliable association between tracklets and detections, we also propose a novel online learning method using an incremental linear discriminant analysis for discriminating the appearances of objects. By exploiting the proposed learning method, tracklet association can be successfully achieved even under severe occlusion. Experiments with challenging public datasets show distinct performance improvement over other batch and online tracking methods.

Subfields: Computer Vision and Pattern Recognition, Electrical and Electronic Engineering, Water Science and Technology

View Paper: https://www.semanticscholar.org/paper/a06d56ba4daafbfa0cf02d1640b840504b2cb089

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

In Electrical and Electronic Engineering

In Water Science and Technology

Citations in the corpus

The Unmanned Aerial Vehicle Benchmark: Object Detection and Tracking (2018, methodology): “Therefore, we evaluate 8 recent algorithms including online methods (CMOT [2], MDP [50], SORT [6] and DSORT [48]) and batch based methods (GOG [35], CEM [30], SMOT [13] and IOUT [7]).”
Simple online and realtime tracking (2016, background): “Many online tracking methods aim to build appearance models of either the individual objects themselves [17, 18, 12] or a global model [19, 11, 4, 5] through online learning.”
3D Multi-Object Tracking: A Baseline and New Evaluation Metrics (2019, methodology): “Modern methods often use the motion model [19], [25], [26] and the appearance feature [25], [27], [28].”
Object Detection from Video Tubelets with Convolutional Neural Networks (2016, unknown): “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 results.”
MOT20: A benchmark for multi object tracking in crowded scenes (2020, methodology): “Following a recent trend [1], [9], [15], we employ two sets of measures that have established themselves in the literature: The CLEAR metrics proposed by Stiefelhagen et al.”
FairMOT: On the Fairness of Detection and Re-identification in Multiple Object Tracking (2020, methodology): “[26] apply Linear Discriminant Analysis to extract reID features for objects which achieves more robust tracking results.”
Tracking the Untrackable: Learning to Track Multiple Cues with Long-Term Dependencies (2017, unknown): “TC ODAL [6] Online 15.”
Memory Based Online Learning of Deep Representations from Video Streams (2017, background): “Recent Multiple Object Tracking algorithms typically adopt appearance and motion cues into an affinity model to estimate and link detections to form tracklets which are afterwards combined into final trajectories [27, 28, 29, 30, 31, 32].”
Multiple Hypothesis Tracking Revisited (2015, background): “In tracklet association, several works [2, 42, 21, 22] train discriminative appearance models of tracklets in order to design a better affinity score function.”
GMMCP tracker: Globally optimal Generalized Maximum Multi Clique problem for multiple object tracking (2015, methodology): “On the other two sequences which no tracking results are reported, we compare our method with competitive approaches which for we have access to their code including KSP [8], DCT [4], CMOT [5], IHTLS [13] and GMCP [27].”
Multiple object tracking: A literature review (2014, unknown): “[44] C Bae & Yoon [164] MATLAB Rezatofighi et al.”
MOTChallenge: A Benchmark for Single-Camera Multiple Target Tracking (2020, methodology): “Following a recent trend (Bae and Yoon 2014; Milan et al. 2014; Wen et al. 2014), we employ three sets of tracking performance measures that have been established in the literature: (i) the frame-to-frame based CLEAR-MOT metrics proposed by Stiefelhagen et al. (2006), (ii) track quality measures…”
Incremental learning algorithms and applications (2016, background): “One major ingredient of such strategies consist in a confidence estimation of the actual model prediction, such as statistical tests, efficient surrogates, or some notion of selfevaluation [8, 43, 78].”
Audio-Visual Speaker Diarization Based on Spatiotemporal Bayesian Fusion (2016, methodology): “Among all these methods, we chose the multiple-person tracker of [37].”
Deep Affinity Network for Multiple Object Tracking (2018, methodology): “The approaches following this pipeline exploit several distinct types of representation models, including; appearance models [16], [17], [18], [19], motion models [20], [21], [22], [23], and composite models [15], [24], [25].”
UA-DETRAC: A new benchmark and protocol for multi-object detection and tracking (2015, methodology): “14 We evaluate the complete MOT systems constructed from com15 binations of ten object tracking schemes (Andriyenko and Schindler, 16 2011; Pirsiavash et al., 2011; Andriyenko et al., 2012; Dicle et al., 17 2013; Wen et al., 2014; Bae and Yoon, 2014; Geiger et al., 2014; 18 Kim et al., 2015; Bochins”
MOTChallenge 2015: Towards a Benchmark for Multi-Target Tracking (2015, methodology): “Robust Online Multi-Object Tracking based on Tracklet Confidence and Online Discriminative Appearance Learning, or TC ODAL [8], is the only online method among the baselines.”