DEVELOPING AN ANDROID APPLICATION FOR ELECTRIC METER RECOGNITION USING DEEP LEARNING

Abstract

In the era of rapid digital transformation, automation in equipment management and identification has become increasingly essential. This thesis, titled “Developing an Android Application for Electric Meter Recognition Using Deep Learning,” focuses on designing and implementing an Android-based system capable of recognizing various types of electric meters using deep learning techniques. The proposed system adopts the YOLO model, a state-of-the-art object detection architecture renowned for its high accuracy and real-time performance. The model is trained on a diverse dataset of electric meters to ensure robust recognition under different real-world conditions, including variations in lighting, viewing angles, and partial occlusions. The developed Android application integrates the trained model to enable realtime electric meter recognition directly through the device camera. Additionally, the system provides corresponding manuals and technical information for each identified meter type, supporting users in operation, troubleshooting, and equipment management. Experimental results demonstrate that the recognition pipeline achieves a high overall accuracy of 93.1%, ensuring fast inference speed and stable performance across multiple Android devices. The study demonstrates a practical and efficient solution for modernizing electric meter identification processes and contributes to ongoing efforts in digital transformation and intelligent system development. Keywords: YOLO, Android application, deep learning