人工智慧於交通運輸潛在應用之回顧

人工智慧的再次興起與科技進步帶動了智慧運輸的發展。產官學界無不努力合作推升國內交通智慧化之程度，國外也不斷產出更精準之模型。如何結合國內外技術應用在國內交通環境是一個重要之在地化課題。本文回顧國內外文獻，首先介紹現有之影像辨識、重疊物件偵測、重識別以及聲音辨識等模式，並提出於交通領域可能之應用參考，其中包含跨攝影機追蹤、高密度車流環境偵測、低能見度環境下之偵測與高噪音機動車輛偵測。提出技術同時也思考應用於國內混合車流環境之可行性。期望透過本文激盪出更多智慧運輸之應用。

關鍵字

物件偵測；影像偵測；聲音偵測；智慧運輸

並列摘要

The recuring rise of artificial intelligence and its progress drive the development of advanced transportation systems. Both the academia and the industry devoted themselves for smarter transportation in Taiwan. Models advancing the frontier of artificial intelligence has been proposed every year. How to adapt these approaches in Taiwan is a critical challenge. This article reviews recent papers where we first introduce existing image recognition, overlapping object detection, re-identification, and audio recognition models. Then, potential applications are proposed, including cross-camera tracking, high-density traffic tracking, low visibility detection, and noisy vehicle detection. We also consider the feasibility of mixed traffic flows in Taiwan. We hope that this article sparks more ideas on applications of intelligent transportation.