Automated Detection of Dysgraphia Symptoms In Primary and Middle School Children

Dysgraphia, a learning disability impacts hand-writing fluency and legibility, which significantly hinder's a child's academic progress and self-esteem. The proposed work utilizes pattern recognition techniques, employing preprocessing techniques like Canny Edge Detection, WaterShed Segmentation and various image enhancement techniques to analyze written characters and extract relevant features using character-wise bounding boxes. This detailed analysis allows the system to capture subtle variations in letter formation and spacing that are often indicative of dysgraphia. Furthermore, A MobileNet architecture has been proposed to assess the system's performance and classify dysgraphia risk levels into three categories: Below Average(BA), Average(AV) and Above Average(AA). This categorization system provides valuable insights for educators and parents, enabling early intervention and tailored support for students struggling with dysgraphia. The model was trained using a data set of 940 images of handwriting samples from students studying in grades one to seven. Students were asked to write 20 words through dictation and another 20 words copied from the board. The proposed system aims to demonstrate immediate results, highlighting its potential for early detection and intervention.