無人機輔助物體偵測與追蹤技術在科技執法中的應用

Abstract

本研究旨在通過擴大覆蓋範圍和即時監控來提升道路安全，探討無人機輔助的超速執法應用。 所提出的系統包括幾個關鍵模組：車輛偵測與多目標追蹤、基於車道線段的速度估測方法，以及車牌辨識。為了確保系統的穩健性和準確性，創建了模擬數據以評估速度估測算法的準確性。此外，還創建了一個臺灣車牌的合成數據集，用於訓練和辨識臺灣車牌。另外，收集無人機數據以評估速度估測方法和車牌辨識的準確性。這個系統基於無人機捕捉圖像，並利用自收集數據進行評估，解決了無人機技術在速度限制執法中的挑戰，旨在確保有效部署並顯著促進整體道路安全的改善。

Aiming to enhance road safety through broader coverage and real-time monitoring, this study investigates the implementation of drone-assisted speed law enforcement applications. The proposed system comprises several key modules: detection and multi-object tracking of vehicles, a lane-line segment-based method for speed estimation, and license plate recognition. To ensure the robustness and accuracy of the system, Simulated data is generated to evaluate the speed estimation algorithms. Additionally, a synthetic dataset of Taiwanese license plates is developed for training recognition models. Drone data is collected to evaluate speed estimation methodologies and the accuracy of license plate recognition. This comprehensive system, based on drone-captured images and utilizing self-collected data for evaluation, addresses the challenges associated with drone technology in speed limit enforcement, ensuring effective deployment and significantly contributing to overall road safety improvements.