Printed Gujarati Text Detection and Recognition (PGTDR)

India is the land of many historic manuscripts and parchments. More than one-third Indian manuscripts are written in Gujarati, according to statistics from the National Mission for Manuscripts. Since handwritten data on paper is difficult to maintain, digitizing these manuscripts and storing them in a database proves helpful in their preservation. The research will use machine learning models trained on a custom dataset to recognize Gujarati text and convert it into a machine-readable format. The research presents an effective, robust and high-precision system to conveniently digitize these manuscripts, using YOLOv8 for character detection with 98% precision and DenseNet model for text recognition with 99.2% accuracy. The system is also capable of detecting text blocks in newspaper articles and this proof of concept will be extremely efficient for detection of texts in Gujarati manuscripts.