ỨNG DỤNG MẠNG YOLOV8 TRONG PHÁT HIỆN HƯ HỎNG TRÊN MẶT ĐƯỜNG BỘ=TOWARDS ROAD DAMAGE DETECTION BY USING YOLOV8

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Duy Trường Giang Nguyễn, Hữu Tuân Nguyễn

Ngôn ngữ: vie

Ký hiệu phân loại:

Thông tin xuất bản: Tạp chí Khoa học Công nghệ Hàng hải, 2024

Mô tả vật lý: tr.73

Bộ sưu tập: Metadata

ID: 479757

Thêm vào giỏ

Các hư hỏng trên mặt đường bộ (ổ gà, ổ trâu, các vết nứt, hỏng - tiếng Anh gọi chung là potholes) là một trong các nguyên nhân chủ yếu gây ra tai nạn giao thông, thiệt hại đối với con đường và các phương tiện giao thông. Việc phát hiện kịp thời và chính xác để từ đó có các biện pháp tu sửa, nâng cấp các hư hỏng này sẽ góp phần giảm tai nạn giao thông và thiệt hại đối với hệ thống đường bộ và phương tiện giao thông. Điều này cũng giúp các nhà quản lý lập kế hoạch bảo dưỡng, bảo trì đường hộ hợp lý và tối ưu hơn. Tuy nhiên, việc phát hiện các hư hỏng trên mặt đường bộ một cách thủ công là một công việc nhàm chán, mất thời gian và không phải lúc nào cũng khả thi, kịp thời. Trong bài báo này, nhóm tác giả đề xuất một giải pháp cho bài toán phát hiện hư hỏng trên mặt đường bộ bằng cách áp dụng mô hình mạng học sâu YOLOv8 trên dữ liệu video thu được từ camera giám sát của con đường cần đánh giá. Kết quả huấn luyện trên một tập dữ liệu hình ảnh gồm 19074 ảnh cho thấy hệ thống đề xuất có độ chính xác huấn luyện tốt (độ chính xác trung bình - mean Average Precision mAP@0.5 là 87,5% với mô hình YOLOv8x sau 200 epoch huấn luyện). Kết quả thực nghiệm khi sử dụng mô hình đã huấn luyện với dữ liệu mặt đường thu thập từ thực tế cũng cho thấy cách tiếp cận sử dụng mạng YOLOv8 cho độ chính xác cao và có khả năng áp dụng trong thực tế.Damages on roads (small holes, large holes, and cracks etc. - potholes in general) are one of the main cause to traffic accidents, disaters and vehices’ devastations. Timely and accurate detection of these potholes and then repair them will help to reduce accidents and damages to roads and vehicles together. This also provides good supports for managers to plan road maintenance and repair more appropriately and optimally. However, manually detecting and measuring road surface damage is a tedious, time-consuming task that is not always feasible and timely. In this article, we propose a solution to the problem of detecting damage on road surfaces (potholes) by applying the YOLOv8 deep learning network model on video data obtained from surveillance cameras. Training results on an image dataset of 19074 images show that the proposed system has good training accuracy (mAP@0.5 is 87.5% with the YOLOv8x model after 200 training epochs). Experimental results of the trained model upon real-life road videos also show that the approach using the YOLOv8 network has high accuracy and is applicable in practice.

1. Đọc trực tuyến

1. Đọc trực tuyến

Tạo bộ sưu tập với mã QR