Abstract:Multi-object tracking (MOT) in the scenario of low-frame-rate videos is a promising solution to better meet the computing, storage, and transmitting bandwidth resource constraints of edge devices. Tracking with a low frame rate poses particular challenges in the association stage as objects in two successive frames typically exhibit much quicker variations in locations, velocities, appearances, and visibilities than those in normal frame rates. In this paper, we observe severe performance degeneration of many existing association strategies caused by such variations. Though optical-flow-based methods like CenterTrack can handle the large displacement to some extent due to their large receptive field, the temporally local nature makes them fail to give reliable displacement estimations of objects that newly appear in the current frame (i.e., not visible in the previous frame). To overcome the local nature of optical-flow-based methods, we propose an online tracking method by extending the CenterTrack architecture with a new head, named APP, to recognize unreliable displacement estimations. Further, to capture the fine-grained and private unreliability of each displacement estimation, we extend the binary APP predictions to displacement uncertainties. To this end, we reformulate the displacement estimation task via Bayesian deep learning tools. With APP predictions, we propose to conduct association in a multi-stage manner where vision cues or historical motion cues are leveraged in the corresponding stage. By rethinking the commonly used bipartite matching algorithms, we equip the proposed multi-stage association policy with a hybrid matching strategy conditioned on displacement uncertainties. Our method shows robustness in preserving identities in low-frame-rate video sequences. Experimental results on public datasets in various low-frame-rate settings demonstrate the advantages of the proposed method.

To Track or to Detect? an Ensemble Framework for Optimal Selection

Mean Shift Tracking Based on Ensemble Multiple Instance Learning

Robust Object Tracking with a Hierarchical Ensemble Framework

APPTracker Plus : Displacement Uncertainty for Occlusion Handling in Low-Frame-Rate Multiple Object Tracking

Global Tracking via Ensemble of Local Trackers

Beyond Local Search: Tracking Objects Everywhere with Instance-Specific Proposals

Robust Tracking-by-detection Using a Selection and Completion Mechanism

Embedding Ensemble Tracking in a Stochastic Framework for Robust Object Tracking

Symbiotic Tracker Ensemble Toward A Unified Tracking Framework

Visual Tracking Via An Ensemble Of Random Classifiers

Object-Level Pseudo-3D Lifting for Distance-Aware Tracking

Ensemble Tracking Based on Diverse Collaborative Framework With Multi-Cue Dynamic Fusion

Tracklet Association in Detect-then-track Paradigm for Long-term Multi-Person Tracking (Extended Abstract)

Multi-target Tracking by Learning Local-to-global Trajectory Models

Distillation, Ensemble and Selection for building a Better and Faster Siamese based Tracker

UTrack: Multi-Object Tracking with Uncertain Detections

Multi-object Tracking by Expanding Long-Tracklets

Multi-object Tracking Through Occlusions by Local Tracklets Filtering and Global Tracklets Association with Detection Responses

Learning to track multiple targets

Object tracking algorithm based on system ensemble

Object Tracking with Joint Optimization of Representation and Classification