Digitizing paper forms with mobile imaging technologies

Abstract

In low-resource settings in developing countries, most records are still captured and maintained using paper forms. Despite a recent proliferation of digital data collection systems, paper forms remain a trusted, low-cost and ubiquitous medium that will continue to be utilized in these communities for years to come. However, it can be challenging to aggregate, share, and analyze the data collected using paper forms. This paper presents mScan, a mobile smartphone application that uses computer vision to capture data from paper forms that use a multiple choice or bubble format. The initial mScan implementation targets the task of digitizing paper forms used to record vaccine statistics in rural health centers in Mozambique. We have evaluated the accuracy and performance of mScan under a variety of different environmental conditions, and our results show that mScan is a robust tool that is capable of accurately capturing and digitizing data from paper forms.