This Repo holds links for all the repositories where the code developed for the completion of the Project of Amrita Space: Advanced CMS Enterprise Powered by a Deep Learning Model resides.
This description provides the need and design of Amrita Space which is made to be an in-house Content Management System. This Content Management brings together problems, the students in Amrita are facing because of lack of a centralized repository of authorized content like exam papers, answer keys, answer sheets, books, PPT(s), lab records, research papers etc. This description also provides an analysis of Optical Character Recognition (OCR) techniques, with a specific focus on the bidirectional LSTM model, in transforming printed or handwritten text into digital formats, with applications spanning document digitization and text analysis that is another core component of the Project. We stress on creating a two-vertical Content Management System for teachers and students respectively, who are the primary users of the app and provide an over-the-top secure application, tested and validated by a cloud service provider. To facilitate the conversion of handwritten text to a digitized document, and we assess the performance of these OCR methods using a diverse dataset, considering metrics such as accuracy, speed, and adaptability to various text types and languages. Our comparative analysis sheds light on the strengths and weaknesses of each approach, providing valuable insights into their practicality in real-world scenarios. In the realm of information retrieval, an innovative approach involves harnessing the power of bidirectional LSTM for semantic search which capitalizes on deep neural networks to unravel the subtleties of textual content, enabling a profound comprehension of its meaning. Bidirectional LSTM, with its ability to grasp text in a manner resembling human language. Noticing the failure of these models on a bundle of images when used via a PDF and high Character Error Rate (CER) we migrate to the profound use of Cloud Vision API, that was a pet project of Alphabet Inc. and layer this over another strip to be used by the teachers and students so that the CMS transcends the boundaries that might be set in a real world sceanio.
The code for "Amrita Space CMS Front-End" is available at AmritaSpaceFront.
The code for "Amrita Space CMS Back-End" is available at AmritaSpaceBack.
The r "Hosted OCR Engine" is available at Hosted OCR Engine.
The code for "Amrita Space Deep Learning Models" is available at AmritaSpaceModels.
The code for "OCR-Engine" is available at OCR-Engine.