Receipt Image Extraction in Budgeting Mobile Application

Money is spent and received almost daily. In most cases, receipts are issued after making a transaction. Budgeting mobile applications let users record these transactions for the purpose of tracking and managing their budget and money flow. The normal budgeting application let users input their transactions manually. In this paper, we proposed an improved feature: computer vision to capture the receipts and extract the data from the receipts.The technology used to extract the data of receipt is Microsoft Azure Form Recognizer which is a specialized Optical Character Recognizer (OCR) service that helps to extract form-like data. The user is required to take a picture of a receipt then the data of the receipt would be extracted. This would help to save the time for the user to repeat the steps to record the transaction by entering the details such as name, amount, date and so on one by one. The tools used for development are Android Studio and Visual Studio Code. Azure Services including Azure Web App Service, Azure Database for MySQL flexible server and Azure Form Recognizer are utilized in this project to develop a web-hosting budgeting mobile application. After the development of the application is done, the real application was tested by 30 respondents and received positive feedback from the respondents where 86.7% of the respondents agreed that the receipt scanning feature and its ability to automatically fill in the transaction details were satisfactory.This budgeting mobile app has the potential to reduce the mistake through manual keying in.