Parking Space Availability Detections from Two Overlapping Cameras Using YOLOv5 and Image Stitching Methods

Misbachul Falach Asy’ari, Chastine Fatichah*, Nanik Suciati

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Capturing a large parking lot's entire area requires the installation of more than one camera. In a parking availability detection system, it is crucial to identify the overlapping area recorded by two cameras. This study aims to identify the overlapping area on two cameras using the image stitching method. The you only look once version 5 (YOLOv5) method is then used to determine whether parking spaces on the stitched image are empty or occupied by cars. The experiments using six YOLOv5 configurations and three different image stitching methods showed that the system could detect the availability of parking slots with the best mean average precision (mAP) score of 0.953. The total number of parking spaces before and after stitching is also compared in this study to demonstrate the accuracy of the number of overlapping parking slots compared to the actual number.

Original languageEnglish
Pages (from-to)276-288
Number of pages13
JournalInternational Journal of Intelligent Engineering and Systems
Volume16
Issue number4
DOIs
Publication statusPublished - 2023

Keywords

  • Image stitching
  • Overlapping camera
  • Parking space detection
  • Smart city
  • YOLO

Fingerprint

Dive into the research topics of 'Parking Space Availability Detections from Two Overlapping Cameras Using YOLOv5 and Image Stitching Methods'. Together they form a unique fingerprint.

Cite this