This paper presents an innovative approach to the detection and segmentation of cracks in civil infrastructure. It can be applied to Unmanned Aerial Vehicles (UAVs) to gather image data, which are then analyzed by a separate system utilizing a Deep Learning approach based on You Only Look Once Version 8 (YOLOvSl-seg) object detection model. The primary goal of this approach is to automate the crack detection process, thereby enhancing productivity, reducing errors, and optimizing cost-effectiveness. The system is trained using a Crack Dataset and employs transfer learning with YOLO VS, sample matching, and loss functions to enhance its performance. Although initially designed for civil infrastructure maintenance, the system's potential applications extend to the broader field of the Internet of Things (IoT), offering the possibility to revolutionize infrastructure inspections.

Concrete Crack Detection and Segmentation in Civil Infrastructures Using UAVs and Deep Learning

Neyestani A.;Ahmed I.;Daponte P.;De Vito L.
2023-01-01

Abstract

This paper presents an innovative approach to the detection and segmentation of cracks in civil infrastructure. It can be applied to Unmanned Aerial Vehicles (UAVs) to gather image data, which are then analyzed by a separate system utilizing a Deep Learning approach based on You Only Look Once Version 8 (YOLOvSl-seg) object detection model. The primary goal of this approach is to automate the crack detection process, thereby enhancing productivity, reducing errors, and optimizing cost-effectiveness. The system is trained using a Crack Dataset and employs transfer learning with YOLO VS, sample matching, and loss functions to enhance its performance. Although initially designed for civil infrastructure maintenance, the system's potential applications extend to the broader field of the Internet of Things (IoT), offering the possibility to revolutionize infrastructure inspections.
2023
979-8-3503-6941-0
Crack Detection
Deep Learning
Drones
UAV
YOLO
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12070/62821
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact