Confidence Trigger Detection: Accelerating Real-time Tracking-by-detection Systems

TL;DR

This paper introduces Confidence-Triggered Detection (CTD), a novel method that accelerates real-time tracking-by-detection systems by selectively skipping detection frames based on tracker confidence, maintaining accuracy while improving speed.

Contribution

The paper presents a new confidence-based approach to selectively bypass detection, significantly enhancing tracking speed without sacrificing accuracy in real-time systems.

Findings

CTD improves tracking speed while maintaining accuracy.

Optimal confidence thresholds balance speed and accuracy.

CTD is robust across various detection models.

Abstract

Real-time object tracking necessitates a delicate balance between speed and accuracy, a challenge exacerbated by the computational demands of deep learning methods. In this paper, we propose Confidence-Triggered Detection (CTD), an innovative approach that strategically bypasses object detection for frames closely resembling intermediate states, leveraging tracker confidence scores. CTD not only enhances tracking speed but also preserves accuracy, surpassing existing tracking algorithms. Through extensive evaluation across various tracker confidence thresholds, we identify an optimal trade-off between tracking speed and accuracy, providing crucial insights for parameter fine-tuning and enhancing CTD's practicality in real-world scenarios. Our experiments across diverse detection models underscore the robustness and versatility of the CTD framework, demonstrating its potential to enable real-time tracking in resource-constrained environments.