LITE: A Paradigm Shift in Multi-Object Tracking with Efficient ReID Feature Integration

TL;DR

LITE introduces a real-time, efficient multi-object tracking paradigm that integrates appearance features directly into the tracking pipeline, significantly improving speed and maintaining accuracy without additional training or inference costs.

Contribution

It presents a novel integrated approach that eliminates extra ReID model training and inference, enhancing speed and simplicity in multi-object tracking.

Findings

Achieves 43.03% HOTA score on MOT17 at 28.3 FPS

Doubles DeepSORT's speed on MOT17

Quadruples speed on MOT20 dataset

Abstract

The Lightweight Integrated Tracking-Feature Extraction (LITE) paradigm is introduced as a novel multi-object tracking (MOT) approach. It enhances ReID-based trackers by eliminating inference, pre-processing, post-processing, and ReID model training costs. LITE uses real-time appearance features without compromising speed. By integrating appearance feature extraction directly into the tracking pipeline using standard CNN-based detectors such as YOLOv8m, LITE demonstrates significant performance improvements. The simplest implementation of LITE on top of classic DeepSORT achieves a HOTA score of 43.03% at 28.3 FPS on the MOT17 benchmark, making it twice as fast as DeepSORT on MOT17 and four times faster on the more crowded MOT20 dataset, while maintaining similar accuracy. Additionally, a new evaluation framework for tracking-by-detection approaches reveals that conventional trackers like DeepSORT remain competitive with modern state-of-the-art trackers when evaluated under fair conditions. The code will be available post-publication at https://github.com/Jumabek/LITE.