Pedagogy without Borders: New Frontiers of Teaching and Technology

Immediate Feedback Assessment Technique® (IFAT®) provides opportunities for students to answer a multiple-choice question till the correct answer is revealed. This process is claimed to promote self-learning while testing, less test anxiety and provide teachers with partial knowledge recognition. Scratch cards-based IFAT® was developed by Epstein and made available to the teaching community through a commercial channel. Despite several advantages, the application of IFAT cards for a final exam of a typical first year physics course requires several manual hours. To ease this tedious process, we have developed software for automatic grading of final exams conducted using scratch card-based IFAT®. This involves scanning all the IFAT cards and developing a pattern recognition software to automatically grade a scratch card-based IFAT ® exam.  

The fundamental principle in our design for automated IFAT ® grading software was to develop an application with features such as fast, efficient, accurate, reliable, scalable, platform agnostic, fulfilling user defined goals, and being intuitive and easy to learn. Java programming language and the Open CV image processing library were chosen to build the software. The architectural design patterns were implemented through the model-view-controller (MVC) design pattern. The MVC pattern also enables effective use of data abstraction and encapsulation, two of the core object-oriented principles. To achieve the accuracy and scalability functional goals, a purpose-built algorithm was constructed to perform the main image processing task of the application. Through various image processing techniques and careful pixel data analysis, the constructed algorithm was capable of detecting (i) scratched, (ii) unscratched and (iii) varying degrees of partially scratched boxes with high accuracy to create a logical representation of the IFAT card. The application also makes use of image registration to properly align the IFAT cards in preparation for pixel analysis. The software in its current form, is being tested for accuracy. Previous measurements demonstrated the capability of high degree of accuracy (> 99%).

Davian Todd ¹, Sergei Chernitsyn ¹, Jun Haong ¹, Ruth Nwankwo ¹, Sindhuja Suresh ² , Omar Alam ¹, Balaji Subramanian^2
1Department of Computer Science, ²Department of Physics and Astronomy,  Trent University, 1600 West Bank Drive, Peterborough, ON, K9J 7B8 CANADA.