Abstract:
Programming assignments are a popular assessment tool in Computer Science education. The value of such assignments depends on how fast and detailed formative feedback is given. This is often a problem due to large class sizes and the cost and difficulty of finding qualified markers. Research on automatic assessment and feedback tools has concentrated on code analysis and correct usage of programming constructs such as loops, conditional statements and recursions. Little research has been done on the automatic analysis of Computer Graphics assignments, where the output is an image rather than text, and problems often have numerous correct solutions. In this paper we present a novel tool for the automatic assessment of Computer Graphics assignments programmed using OpenGL. Our tool uses invariants and intermediate data obtained from the OpenGL state machine and rendering pipeline. We evaluated our tool using assignments from a large year 3 Computer Graphics class and found that the majority of errors found by human markers were correctly identified by our application.