Segmentation and Contour Detection for handwritten mathematical expressions using OpenCV

Contour Detection owes its own significance in performing semantic segmentation and object classification. It is possible to recognize the edges of objects and readily locate them inside an image by employing contour detection. It is frequently the first step or achievable milestone in a varied range of exciting applications, such as extracting the foreground object from an image, classical-image segmentation methodology, identification or detection of the object, and many more. In this study, we have endeavored to perform contour detection using the OpenCV Adaptive Threshold method on the Aida handwriting math recognition dataset, containing ten batches of 10K images. Our experimentation successfully segmented the inputted mathematical expressions with an accuracy of 94.3% on the acquired dataset.