Optimal AI Model for Industrial AR Application based on Object Detection
Kanuengnij Kubola, Bhusana Kongon, Prawit Boonmee, Prajaks Jitngernmadan · 2023
In the modern industries and manufacturing techniques, quick and precise maintenance is a vital concept. Several approaches fulfill these requirements, including the Augmented Reality (AR) technology. The AR technology, combined with real-time object detection techniques, offers a possible solution to improve and disrupt the maintenance techniques in the near future. For finding the most suitable AI model for this purpose, this study compared the pre-trained YOLOv5, Resnet50, and SSD MobileNet models in detecting PLCnext components. Using 960 images for training and testing purpose, the models were trained and tested, highlighting their real-time capabilities. The results show that the pre-trained YOLOv5 model is the best to enhance interactive monitoring, minimize downtime, and boost efficiency in industrial settings and suitable for AR application development in automation 4.0 after being trained with required images.