Deep Neural Network for Recognition of Enlarged Mathematical Corpus
Sakshi Sakshi, Sachin Lodhi, Vinay Kukreja · 2022 International Conference on Decision Aid Sciences and Applications (DASA) · 2022
Recent works influenced by the robust and outstanding performance of deep learning-based recognition models show the less attended research on an enlarged corpus of handwritten mathematical text. This article focuses on proposing a deep neural network-based model that has been processed and trained for a real-time dataset of mathematical expressions collected from school and university students of Punjab and Madhya Pradesh states of India. Mathematical expressions are part of education and knowledge-based science learning. Thus, predicting them on their handwritten source using a neural network is the novel approach of the article. A total of 236057 images have been obtained after segmentation, and the sample data has been collected from 1000 users. The proposed deep neural model exclusive of three dense layers (TrioNet) gives us an accuracy of 99.1% while working with 20 epochs.