Faster object tracking pipeline for real time tracking

11/08/2020
by   Parthesh Soni, et al.
0

Multi-object tracking (MOT) is a challenging practical problem for vision based applications. Most recent approaches for MOT use precomputed detections from models such as Faster RCNN, performing fine-tuning of bounding boxes and association in subsequent phases. However, this is not suitable for actual industrial applications due to unavailability of detections upfront. In their recent work, Wang et al. proposed a tracking pipeline that uses a Joint detection and embedding model and performs target localization and association in realtime. Upon investigating the tracking by detection paradigm, we find that the tracking pipeline can be made faster by performing localization and association tasks parallely with model prediction. This, and other computational optimizations such as using mixed precision model and performing batchwise detection result in a speed-up of the tracking pipeline by 57.8% (19 FPS to 30 FPS) on FullHD resolution. Moreover, the speed is independent of the object density in image sequence. The main contribution of this paper is showcasing a generic pipeline which can be used to speed up detection based object tracking methods. We also reviewed different batch sizes for optimal performance, taking into consideration GPU memory usage and speed.

READ FULL TEXT

page 2

page 9

research
03/21/2023

OmniTracker: Unifying Object Tracking by Tracking-with-Detection

Object tracking (OT) aims to estimate the positions of target objects in...
research
09/27/2019

Towards Real-Time Multi-Object Tracking

Modern multiple object tracking (MOT) systems usually follow the trackin...
research
07/23/2014

FollowMe: Efficient Online Min-Cost Flow Tracking with Bounded Memory and Computation

One of the most popular approaches to multi-target tracking is tracking-...
research
03/23/2021

Global Correlation Network: End-to-End Joint Multi-Object Detection and Tracking

Multi-object tracking (MOT) has made great progress in recent years, but...
research
03/18/2021

Real-Time Visual Object Tracking via Few-Shot Learning

Visual Object Tracking (VOT) can be seen as an extended task of Few-Shot...
research
08/22/2023

TrackFlow: Multi-Object Tracking with Normalizing Flows

The field of multi-object tracking has recently seen a renewed interest ...

Please sign up or login with your details

Forgot password? Click here to reset