Advanced search
Start date

Automation of the scanning, mapping and identification processes during the grading exams process


PODD is a cloud software to grade exams of high schools, universities, among others. Besides multiplatform (with desktop applications, tablets or mobile), the system supports the grading of assessments scanned from paper both objective questions (multiple choice) questions and simple (one-word answers or V / F), and aims to provide contextual analysis and visualization tools to support teacher for grading of complex (written long answers) questions. Teacher (user) using an acquisition device for image / video transforms physical assessments into images. In the case of acquisition devices in video format (such as a mobile phone, tablet or webcam) scanning will be aided by an automatic process of choosing the best image. The scanned pages are sent to the cloud for correction.Computer vision processes are performed for identification and standardization of evaluations. These processes locate the uniqueness and the positions of each question on the sheet as well as the type of question (objective, dissertation, V / F, etc.) and the value of its score. This step makes it unnecessary to use the fiducial markers (QR-code, various marks such as triangles, etc.) in the evaluation sheet. This makes the process more flexible for the user who can keep doing the race the way you're used to.The next step concerns the list of students with the evidence sent to PODD. This process can be performed as in the previous step with the spaces of evidence likely to be regions with the student's name, eg handwritten in the header were identified. Then, with the integration of technologies such as Intelligent Character recogntion (ICR), CAPTCHA and the Mechanical Turk PODD can automatically make the association of each student with his test. Handwritten understanding word through these technologies cost-tenths of cents, so can be considered a little expressive value in helping scale.With the issues identified and students associated modules correction begin to run:. (1) automatic correction module simple objective questions and written responses and (2) support module for correction of long written answers. Finalized the process, it is given the possibility to the user to do the validation of all such content.After this, teacher (user) can involve other teachers (or monitors) and send different packets of evidence to be corrected by each. The screen correction of written questions long been designed to be simple and practical. Completed the semi-automatic fixes, all data are compiled and the final result is generated. Various reports and views of the class can be performed. In addition, feedback from brokers on long written questions may be sent by email to the student to observe their mistakes and successes. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Please report errors in scientific publications list by writing to: