An Investigation of Novel Combined Features for a Handwritten Short Answer Assessment System

Abstract

This paper proposes an off-line automatic assessment system utilising novel combined feature extraction techniques. The proposed feature extraction techniques are 1) the proposed Water Reservoir, Loop, Modified Direction and Gaussian Grid Feature (WRL_MDGGF), 2) the proposed Gravity, Water Reservoir, Loop, Modified Direction and Gaussian Grid Feature (G_WRL_MDGGF). The proposed feature extraction techniques together with their original features and other combined feature extraction techniques were employed in an investigation of the efficiency of feature extraction techniques on an automatic off-line short answer assessment system. The proposed system utilised two classifiers namely, artificial neural networks and Support Vector Machines (SVMs), two type of datasets and two different thresholds in this investigation. Promising recognition rates of 94.85% and 94.88% were obtained when the proposed WRL_MDGGF and G_WRL_MDGGF were employed, respectively, using SVMs.

ICFHR 2020 Competition on Short answer ASsessment and Thai student SIGnature and Name COMponents Recognition and Verification (SASIGCOM 2020)

“…Examples of genuine and skilfully forged Thai name components can be seen in Figure 3. Short answer dataset [11]: There are 104 exam papers in this dataset, 52 of which were written using cursive handwriting; the rest of 52 papers were written with printed handwriting. This dataset was used for the word spotting task, where known words (corresponding to the correct answers) should be spotted then marked.…”