This repository is dedicated to implementing Deep Learning-based Scene Text Recognition models, utilizing a two-step approach involving Text Detection (TD) and Text Recognition (TR). The TD step employs YOLOv8, while the TR step utilizes a Convolutional Recurrent Neural Network (CRNN).

YOLOv8-CRNN Scene Text Recognition

The project workflow is straightforward: Given an image, text detection and recognition are performed through YOLOv8 and CRNN models, respectively. The process involves the detection and extraction of texts using YOLOv8, storing the resulting texts as a collection of cropped text images. These cropped images serve as input for the CRNN model, which recognizes all the text within them. The final results are then plotted on the original image.

You can see more details about the repo in this link