Answers are grounded in this paper and cite their sources.
In Karnataka, Kannada is a commonly spoken official state language. The handwritten historical Kannada characters are preserved in the archaeological department of the manuscript preservation center, but these documents degrade the nature of the document and are also complex to read and understand the content presented in the document. So, it is highly needed for digitalizing the handwritten Kannada character files and identifying their originality. Pattern recognition in the handwritten Kannada character recognition model is a challenging research work based on its complex structure and large-scale vocabulary. The images are acquired from the standard benchmark dataset in the initial phase. Then, the images are given to the ensemble model for character recognition to improve reliability and robustness. Here, the ensemble model Weighted Averaging Score-based Ensemble Network (WAS-EnsembleNet) is designed by incorporating the Multi-scale Deep Convolutional Neural Network (M-DCNN), MobileNet, and Residual Attention Network (RAN), where the parameters in these models are optimized with the help of Hybrid algorithm Dolphin Echolocation-based Lion Optimization Algorithm (DE-LOA). Then, the Weighted Average score-based Classification is performed by getting the predictive score from these techniques. Finally, classified outcomes are attained. The efficacy of the recommended Kannada handwritten recognition model is compared by diverse performance measures.
No takes yet. Share an insight, caveat, or question.
Parashivamurthy et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: