TY - JOUR
T1 - AI and IoT-based Automated Camera Maintenance System
T2 - A Study of Improving Production Efficiency and Safety through Predictive Maintenance
AU - Jun, Jihwan
AU - Kim, Tae Yong
AU - Lee, Jieun
AU - Jin, Taeheon
AU - Park, Seungmin
AU - Jeong, Jongpil
N1 - Publisher Copyright:
© 2025, World Scientific and Engineering Academy and Society. All rights reserved.
PY - 2025
Y1 - 2025
N2 - Traditional camera maintenance methods rely on periodic manual inspections, which are time consuming, costly, and inherently limited in that they only allow problems to be detected and addressed after they occur. In large-scale industrial sites operating hundreds of cameras, such methods frequently lead to human errors and delays in response, resulting in decreased safety and productivity. This study proposes an automated and intelligent predictive maintenance system by integrating artificial intelligence (AI)-based video analysis technology with industrial Internet of Things (IIoT) systems to overcome these limitations. The proposed system collects highresolution video data in real-time and operates through a series of processes including image preprocessing, feature extraction, anomaly detection, severity classification, and maintenance alert transmission. Utilizing CNN-based deep learning algorithms and OpenCV image processing techniques, the system can automatically detect issues such as lens contamination, focus blur, and image degradation. When anomalies are identified, they are immediately classified, and alerts are sent in real-time via a cloudbased notification system. Additionally, maintenance history is automatically logged and analyzed in a database, supporting the development of long-term asset management strategies. Experimental results in real industrial environments demonstrate that the proposed system improves detection accuracy by over 90–95% compared to manual inspection methods, reduces alert response time to within seconds, and lowers maintenance time and costs by more than 70% and 40%, respectively. This research validates the practical effectiveness of automated and predictive maintenance in camera systems as a core technology for smart factory implementation and is expected to contribute to the development of more scalable maintenance frameworks through integration of multi-sensor data and further advancements in predictive algorithms. Index Terms—Predictive Maintenance, Industrial IoT (IIoT), Automated Camera, AI Image Analysis, SmartFactory.
AB - Traditional camera maintenance methods rely on periodic manual inspections, which are time consuming, costly, and inherently limited in that they only allow problems to be detected and addressed after they occur. In large-scale industrial sites operating hundreds of cameras, such methods frequently lead to human errors and delays in response, resulting in decreased safety and productivity. This study proposes an automated and intelligent predictive maintenance system by integrating artificial intelligence (AI)-based video analysis technology with industrial Internet of Things (IIoT) systems to overcome these limitations. The proposed system collects highresolution video data in real-time and operates through a series of processes including image preprocessing, feature extraction, anomaly detection, severity classification, and maintenance alert transmission. Utilizing CNN-based deep learning algorithms and OpenCV image processing techniques, the system can automatically detect issues such as lens contamination, focus blur, and image degradation. When anomalies are identified, they are immediately classified, and alerts are sent in real-time via a cloudbased notification system. Additionally, maintenance history is automatically logged and analyzed in a database, supporting the development of long-term asset management strategies. Experimental results in real industrial environments demonstrate that the proposed system improves detection accuracy by over 90–95% compared to manual inspection methods, reduces alert response time to within seconds, and lowers maintenance time and costs by more than 70% and 40%, respectively. This research validates the practical effectiveness of automated and predictive maintenance in camera systems as a core technology for smart factory implementation and is expected to contribute to the development of more scalable maintenance frameworks through integration of multi-sensor data and further advancements in predictive algorithms. Index Terms—Predictive Maintenance, Industrial IoT (IIoT), Automated Camera, AI Image Analysis, SmartFactory.
KW - Computer Vision
KW - Deep Learning
KW - Monocular Depth Estimation
KW - Object Detection
KW - RGB-D Fusion
UR - https://www.scopus.com/pages/publications/105019305010
U2 - 10.37394/23207.2025.22.156
DO - 10.37394/23207.2025.22.156
M3 - Article
AN - SCOPUS:105019305010
SN - 1109-9526
VL - 22
SP - 1955
EP - 1970
JO - WSEAS Transactions on Business and Economics
JF - WSEAS Transactions on Business and Economics
ER -