Plataforma mobile para correção de exercícios de alfabetização por imagem com processamento em nuvem

Abstract

Literacy is an essential process in educational formation and the continuous assessment of student performance is essential to ensure the effectiveness of teaching. This work presents the development of a system with a mobile application for the automated correction of handwritten exercises, with a focus on literacy. The goals are to offer a platform that facilitates the correction and evaluation process, providing a quick and accurate feedback to students and teachers, in addition to generating data to be processed in large volumes. The application allows the use of the cell phone’s own camera to record exercises, analyzes the handwritten areas sent by the students, evaluates the answers based on a template previously provided by the teacher, corrects the exercises and records the results in a database. The system architecture was implemented using React-Native as framework for front-end and Firebase as the back-end platform, with the Cloud Firestore database for efficient and secure data storage. The system was designed to comply with the requirements in this defined scope. It was possible to implement it in a decoupled way, enabling the expansion of the current system in a more coherent way. Tests were carried out to evaluate the use of implemented algorithms of Optical Character Recognition (OCR) in 89 images, demonstrating that the best performance was from Google Cloud Vision with 91.88% of accuracy.