Robotic Process Automation in IoT: Enhancing Object Localization Using YOLOv3-Based Class Algorithms

Abstract

A hybrid YOLOv3-Mask RCNN model is suggested in this paper to improve object localization in IoT-enabled Robotic Process Automation (RPA) systems. Accurate and efficient item recognition is essential for automating tasks like logistics, inventory management, and assembly line operations as the Internet of Things continues to permeate many industries. By utilizing both the accurate segmentation provided by Mask-RCNN and the real-time detection capacity of YOLOv3, the hybrid model improves both processing speed and localization accuracy. Experimental results show that the hybrid model performs better than conventional techniques, with a processing time of 35 milliseconds and a precision of 0.92, recall of 0.91, mAP of 0.93, and IoU of 0.88. These measurements demonstrate how well the model works in the kind of dynamic, complicated situations found in Internet of Things applications. This work tackles the problems of varying object sizes, orientations, and partial occlusions by offering a strong framework for object localization. These results highlight the possibility of applying hybrid deep learning models in real-world Internet of Things situations, improving the effectiveness and dependability of automated systems in different industries.