為了交通安全,政府推動了許多道路安全規範,包括騎乘機車戴安全帽、禁止駕駛使用行動電話、汽車前後座繫安全帶及嚴懲酒後駕車等措施。國內嚴重交通事故也因此有下降,然而交通事故總數卻呈現上升的現象,如何降低因交通事故導致的醫療成本與生命財產的損失是刻不容緩的。現今有許多人工智慧的產品進入到我們的生活之中,其中影像辨識是非常熱門和重要的一環,可以透過大量的影像資料與數據資料將不同情況分類。本研究使用Github上的開源資料YOLO做為基底,以監視器的資料放進模型之中訓練,結合通訊軟體。透過本研究的辨識方法,可以及早發現車禍是故的發生並通報相關單位,藉以降低車禍受害者在車禍中醫療成本與生命財產的損失。
In recent years, road safety regulations have been strongly promoted. These include the wearing of helmets on motorcycles, banning the use of mobile phones, wearing seat belts in front and rear seats of a vehicle, and strictly prohibit any driving under the influence of alcohol, which have led to a decrease in the number of serious accidents in domestic. Nowadays, many artificial intelligence products have entered people’s daily life, among which image recognition is a very popular and important part, which can classify different situations through a large amount of image data and data. This study uses the open source package YOLO on Github as the base, and the traffic monitor video is put into the model for training and combining with the communication software. Through the identification method of this study, the relevant authorities can be notified as soon as possible when the accident occured, so as to reduce the medical cost and loss of life and property of the victims in the car accident.