Handwriting recognition

Author: Ana Chikashua
Keywords: Handwriting recognition, YoloV8, ResNet, CNN
Annotation:

This project focuses on developing an Optical Character Recognition (OCR) system specifically designed for the Georgian language. The main objective of the project is to create an accurate and efficient OCR model by leveraging the YOLOv8 architecture and training it on a dataset comprising 48625 Georgian language samples. The trained model is then deployed as a web service, allowing users to easily access and utilize its capabilities. To achieve this, the project involves several key steps. Firstly, an extensive dataset of Georgian language samples is collected and curated to ensure diverse representation across various styles, and variations. Next, the YOLOv8 model, known for its effectiveness in object detection and classification tasks, is adapted and retrained using the collected dataset. This process involves adjusting the model's parameters, fine-tuning its weights, and optimizing it for the specific requirements of OCR. Once the model training is complete, a web service is developed to host the OCR system. Users can interact with the system through a user-friendly interface provided by the web service. This interface allows users to write Georgian text, and the OCR model processes the images to extract the text accurately. In addition, model allows users to upload images containing Georgian text, and the OCR model processes the images to extract the text accurately, which is quite problematic. Overall, this project presents a robust and efficient OCR solution specifically tailored for the Georgian language. The utilization of the YOLOv8 architecture, combined with the retraining on a large dataset, contributes to improved accuracy and effectiveness in recognizing Georgian text. By deploying the model as a web service, the system provides a user-friendly platform for easy access and utilization of the OCR capabilities. This project serves as a valuable resource for applications requiring Georgian language text extraction and analysis.