Image-Based Automatic Energy Meter Reading Using Deep Learning

We propose to perform an image-based framework for electrical energy meter reading. Our aim is extract the image region that depicts digits and then recognize them record consumed units. Combining readings of serial numbers units, automatic billing system using Internet Things a graphical user interface deployable in real-time setup. However, such extraction character recognition become challenging due variations caused by several factors as partial occlusion dust on display, orientation scale camera positioning, non-uniform illumination shades. To this end, our work evaluates compares state-of-the art deep learning algorithm You Only Look Once (YOLO ) along with traditional handcrafted features text recognition. dataset contains 10,000 images meters further expanded data augmentation in-plane rotation scaling make algorithms robust these variations. For training evaluation, annotated produce ground truth all images. Consequently, YOLO achieves superior performance over average rate 98% digits. It proves be against mentioned compared features. proposed method can highly instrumental reducing time effort involved current reading, where workers visit door door, take manually from