itype

An automated typing tool

Over 90% of scanning systems produce outputs that are inform of images or Portable Document Format (PDF) that are not editable by most of the openly accessible document processing software. In education sector for instance, due to increased uptake of online education, libraries must convert analog resources to digital formats for ease of online delivery. Document creators not limited to Office administrators , students alike must convert paper-based documents, as well as handwritten documents to digital format. However, most document scanning systems in place do not allow editing of the scanned documents to improve it or make necessary typing corrections.  The editing capability is usually provided as an additional service at a fee which is costly for the public sector and more so, most editing software are provided as either a standalone desktop application or mobile application.  There is therefore need for an openly accessed tool with the capability of editing the scanned output

Our solution itype is an automated typing tool that uses OCR technology to recognize characters in documents (images and pdfs) provided as input by users, types them and readily avails the output in editable format.

Major processes involved are enhancement basically for images to ensure high quality output this is done using OpenCV, text extraction using tesseract ocr and pdf plumber , as well as instantiation of a processed editable and downloadable document using docx

We used django framework, python for backend, html css, javascript and bootstrap for front end, poppler, pdf plumber for pdf handling, pytesseract, docx for docment handling, PIL for image handling.

This is all being powered by git and github for version control

Our solution serves Document Creators that is to say secretaries, librarians, typists, students.

In Uganda, Digital divide is rampant where there is inequity in accesss to computing tools but with increased uptake of smartphones, atleast half of our target market can afford smartphones of which our solution shall be responsive to smartphone screens hence lessen on the task of typing but instead just be required to do simple editings.

The team is comprised of alumni from Muni University and have interfaced as well with this problem, we undertook our bachelors in information systems from 2019-2022 where we were passionate about software development and are working towards creating a positive impact in the society we live in

At the beginning of the year 2022 we visited a secretary's office to hand in one of our class assignments and found her so busy that she had been tasked to retype different sets of already typed work and combine them. On seeing the hardships she was going through, this caught our attention and decided to come up with a solution that gave rise to itype.

We did research were we engaged our target users to determine whether they really need this tool and if they will be able to adopt to it

Some of the document creators in two universities in Arua -Uganda that we interacted with told us that they daily interface with piles of documents some of which need slight corrections but instead they are required to retype the whole document and they were more than willing to adopt to its usage if its pushed to the market.

In our local context its the first of its kind, and a local product being built using the current technology.

We are to deliver it as a non platform specific to suit all users with different gadgets and operating systems. We expect our product to attract many potential users and will make document creators more productive than just typing documents by reducing the time they would have taken to type them.

Its ability to also tackle handwritten characters gives us a plus because handwritten text extraction has not yet been focussed on.

Saving energy that is the lesser the time taken typing the lesser the power consumption of computing devices as well as emissions.

Picking out all other special characters like lines, images.

Rolling out our first version to the market

Its developed using an agile methodology in the django framework where Python is our major. we applied python for back end, html , css, javascript, bootstrap for front end, OpenCV used  for image enhancement that is to say deblurring, fixing image contrast and brightness. 

Pillow being built on top of PIL (Python Image Library) communication between python and the PIL module to add image processing capabilities to the Python interpreter.

Pdf2image a python module to convert portable document formats (pdf) containing text to high quality image formats working hand in hand with Poppler for pdf rendering 

Text extraction using pdf plumber, Tesseract OCR via an installable python binding known as Pytesseract. 

Python docx module together with the docx utility for handling prepared editable format output

We look forward to serving 500 people in the next year

Limited knowledge in Machine Learning for properly developing a handwritten model

Limited capital to finance our project for example GPU computers where the model will be tested when built, as well as financing cloud storage capabilities.

Security constraints that may show up.

None

We aim at providing the best typing service that is affordable to all. 

In this our key resources are Cloud Servers, hosting space with key activities as building a handwritten character recognition model for easier scalability and better  accuracy. Current Stakeholders are Laban Kulubasi and Osuta Denis but we are planning to get

Our intervention is a software as a service whose Customer Segments involve direct beneficiaries who are Document creators not limited to Secretaries, Typists, Students and the general public. Document Processing Institutions will be in position to pay for this service as well

In our Value Proposition, our customers should expect to type documents quicker and more accurate.

We plan to reach our users through a shared Whatsapp group, mailing list for updates regarding improved features.

In our Cost Structure, our biggest expenditure will be on infrastructure for testing our model as well as hosting space. As we scaleup we will handle this by acquiring a GPU computer as local infrastructure.

As a means of sustaining our value proposition as well as gain revenue we plan for a subscription fee for usage of the handwritten text recognition feature, Optional donations

We intend to offer one of our proposed services as handwritten text recognition feature on a subscription basis. On fully developing its model,  testing it with positive results, and pushing it to the market our potential customers will be able to pay for it.