How to Train Your Deep Multi-Object Tracker
Authors:
- 1 June 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 6786-6795
Abstract
The recent trend in vision-based multi-object tracking (MOT) is heading towards leveraging the representational power of deep learning to jointly learn to detect and track objects. However, existing methods train only certain sub-modules using loss functions that often do not correlate with established tracking evaluation measures such as Multi-Object Tracking Accuracy (MOTA) and Precision (MOTP). As these measures are not differentiable, the choice of appropriate loss functions for end-to-end training of multi-object tracking methods is still an open research problem. In this paper, we bridge this gap by proposing a differentiable proxy of MOTA and MOTP, which we combine in a loss function suitable for end-to-end training of deep multi-object trackers. As a key ingredient, we propose a Deep Hungarian Net (DHN) module that approximates the Hungarian matching algorithm. DHN allows estimating the correspondence between object tracks and ground truth objects to compute differentiable proxies of MOTA and MOTP, which are in turn used to optimize deep trackers directly. We experimentally demonstrate that the proposed differentiable framework improves the performance of existing multi-object trackers, and we establish a new state of the art on the MOTChallenge benchmark. Our code is publicly available from https://github.com/yihongXU/deepMOT.Keywords
All Related Versions
This publication has 36 references indexed in Scilit:
- FaceNet: A unified embedding for face recognition and clustering2015
- Data-driven 3D Voxel Patterns for object category recognition2015
- Fast Feature Pyramids for Object DetectionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2014
- A multi-sensor fusion system for moving object detection and tracking in urban driving environments2014
- Multi-target Tracking by Lagrangian Relaxation to Min-cost Network Flow2013
- Continuous Energy Minimization for Multitarget TrackingIEEE Transactions on Pattern Analysis and Machine Intelligence, 2013
- Evaluating Multiple Object Tracking Performance: The CLEAR MOT MetricsEURASIP Journal on Image and Video Processing, 2008
- Histograms of Oriented Gradients for Human Detection2005
- An algorithm for tracking multiple targetsIEEE Transactions on Automatic Control, 1979
- The Hungarian method for the assignment problemNaval Research Logistics Quarterly, 1955