http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
보행교통사고 다발지역 예측을 위한 딥러닝의 적용 : 수도권 노인과 어린이 보행교통사고를 중심으로
전희정(Jun, Hee-Jung),강승엽(Kang, Seungyeoup),정수영(Jung, Suyoung),김태완(Kim, Taewan),조철호(Cho, Cheol-Ho),주원영(Jhoo, Won Young),김지영(Kim, Ji Young),허재필(Heo, Jae-Pil) 대한국토·도시계획학회 2021 國土計劃 Vol.56 No.7
An efficient prediction of the pedestrian crash hot spots of the transportation disadvantaged is important for pedestrian-friendly environments and the promotion of sustainable development. This study aims to predict elderly and child pedestrian hot spots in the Seoul Metropolitan Area. For the empirical analysis, we used the traffic accident analysis system data and collected Google Street View images of elderly and child pedestrian crash hot spots and non-hot spots. Then, we conducted experiments using five deep learning models, including VGG16, VGG19, ResNet50, ResNet101, and InceptionV3. We also employed the CAM analysis to visualize the factors contributing to pedestrian crashes. The empirical analysis showed that the VGG16 model was the best model in predicting elderly and child pedestrian crash hot spots. In addition, the CAM analysis suggested that narrow widths of roads, physical facilities for transportation safety, and low openness were related to the pedestrian safety of the elderly and children.