A monocular rear-view motorcycle warning algorithm based on deep learning

Message:
Article Type:
Research/Original Article (دارای رتبه معتبر)
Abstract:
In Iran, motorcycles are one of the most vulnerable types of vehicles among other road users, with a significant volume of accident statistics in the country. Therefore, in this article, a suitable solution is proposed to reduce car accidents with these motorcycles equipped with black windshields. In this method, a single camera mounted on the side mirror of the driver's assistance is used. This intelligent automatic monitoring can help drivers to pay more attention to the surrounding. This warning algorithm was formed on a combination of motorcycle detection and depth estimation tasks based on deep learning methods. For detecting this category of motorcycles, different models of YOLO algorithms have been compared. According to the speed and accuracy, the fine-tuned YOLOV4 model with an average accuracy of 80% and at a real-time speed of 35 frames per second is used for the detection stage, which has the best performance in comparison to others. This model was fine-tuned based on 2000 images taken from the mentioned motorcycles in Tehran. In the second step, the Monodepth2 model is used to estimate the depth map of a single image. This model was fine-tuned on 6000 stereo images taken from the streets of Tehran by using the MYNT-EYE camera. The contribution of these two algorithms produces a state-of-the-art result for monocular depth estimation, which can estimate the distance of the detected motorcycle with the average error of 36 centimeters at a real-time speed of 20 frames per second.
Language:
Persian
Published:
Journal of Transportation Engineering, Volume:13 Issue: 3, 2022
Pages:
1683 to 1710
magiran.com/p2433910  
دانلود و مطالعه متن این مقاله با یکی از روشهای زیر امکان پذیر است:
اشتراک شخصی
با عضویت و پرداخت آنلاین حق اشتراک یک‌ساله به مبلغ 1,390,000ريال می‌توانید 70 عنوان مطلب دانلود کنید!
اشتراک سازمانی
به کتابخانه دانشگاه یا محل کار خود پیشنهاد کنید تا اشتراک سازمانی این پایگاه را برای دسترسی نامحدود همه کاربران به متن مطالب تهیه نمایند!
توجه!
  • حق عضویت دریافتی صرف حمایت از نشریات عضو و نگهداری، تکمیل و توسعه مگیران می‌شود.
  • پرداخت حق اشتراک و دانلود مقالات اجازه بازنشر آن در سایر رسانه‌های چاپی و دیجیتال را به کاربر نمی‌دهد.
In order to view content subscription is required

Personal subscription
Subscribe magiran.com for 70 € euros via PayPal and download 70 articles during a year.
Organization subscription
Please contact us to subscribe your university or library for unlimited access!