2017/12/28 by Md Zahangir Alom, Alom, Md Zahangir, Peheding Sidike +6
Computer Science · Engineering · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Handwritten Text Recognition Techniques #Natural Language Processing Techniques #Vehicle License Plate Recognition
paper · pdf · doi:10.48550/arxiv.1712.09872
openalex publication_date 2017/12/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In spite of advances in object recognition technology, Handwritten Bangla\nCharacter Recognition (HBCR) remains largely unsolved due to the presence of\nmany ambiguous handwritten characters and excessively cursive Bangla\nhandwritings. Even the best existing recognizers do not lead to satisfactory\nperformance for practical applications related to Bangla character recognition\nand have much lower performance than those developed for English alpha-numeric\ncharacters. To improve the performance of HBCR, we herein present the\napplication of the state-of-the-art Deep Convolutional Neural Networks (DCNN)\nincluding VGG Network, All Convolution Network (All-Conv Net), Network in\nNetwork (NiN), Residual Network, FractalNet, and DenseNet for HBCR. The deep\nlearning approaches have the advantage of extracting and using feature\ninformation, improving the recognition of 2D shapes with a high degree of\ninvariance to translation, scaling and other distortions. We systematically\nevaluated the performance of DCNN models on publicly available Bangla\nhandwritten character dataset called CMATERdb and achieved the superior\nrecognition accuracy when using DCNN models. This improvement would help in\nbuilding an automatic HBCR system for practical applications.\n