Development of road asset mapping using dashcam
Road asset mapping significantly benefits transportation authorities, infrastructure management, and road users. Recent advancements in Geographic Information Systems (GIS) and digital mapping technologies have substantially improved inventory and asset management. However, technologies such as Ligh...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Penerbit UniMAP
2025
|
| Subjects: | |
| Online Access: | https://umpir.ump.edu.my/id/eprint/46554/1/14-29%2B%281480%29%2BDevelopment%2Bof%2BRoad%2BAsset%2BMapping%2Busing%2BDashcam.pdf https://doi.org/10.58915/amci.v14i4.1480 https://umpir.ump.edu.my/id/eprint/46554/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Road asset mapping significantly benefits transportation authorities, infrastructure management, and road users. Recent advancements in Geographic Information Systems (GIS) and digital mapping technologies have substantially improved inventory and asset management. However, technologies such as Light Detection and Ranging (LiDAR) and mobile mapping cameras are costly compared to dashcams. Therefore, this study proposes a cost-effective road asset mapping system by leveraging dashcam video data, object detection using You Only Look Once version 8 (YOLOv8), and GIS for spatial visualization. In this study, images were extracted from dashcam videos using VLC Media Player and processed through an annotation pipeline in Roboflow, where bounding boxes and labels were assigned to road assets such as road signs, streetlights, and traffic lights. The dataset was then divided into training, validation, and test sets for model development. YOLOv8, selected for its high accuracy in object detection and segmentation, was trained to recognize these assets, achieving a precision of 0.895, recall of 0.873, and mean Average Precision (mAP) of 0.876 at 50% Intersection over Union (IoU). To integrate YOLOv8 with GIS, the detected road assets were geotagged based on GPS metadata from the dashcam footage, allowing spatial mapping within a GIS platform. The identified assets were then visualized on a GIS interface, facilitating efficient road asset inventory management. This approach demonstrates that low-cost dashcam-based data collection, combined with AI-powered object detection and GIS mapping, offers a viable alternative to expensive mapping technologies. Future research should focus on enhancing dataset quality and expanding the range of detectable assets to further improve system accuracy and applicability. |
|---|