Automated-Electronic-Meter-Reading-System-using-YOLO-Architectures

● The project consists of 3 major components: Identifying the counter region, localizing and cropping the counter region, and digit recognition. ● Manually collected and annotated 7,877 images of meters for counter detection. ● Utilized YOLOs for counter-detection as well as recognition in a two-stage AMR system. ● Utilized a custom function for the detection of counters and cropping of the counter region, thus making a new dataset of cropped meter images which was annotated for each digit and trained using the customized OCR for digit recognition. ● The system achieved remarkable recognition rates of 99.74% for counters and 88.70% for digit recognition.