Enhancing the Power of CNN Using Data Augmentation Techniques for Odia Handwritten Character Recognition

Table 11

Performance comparison on handwritten NITROHCS v1.0 Odia character databases.

Database	Work reference		Features	Recognition classifier	Recognition accuracy (%)
NITROHCSv1.0 handwritten database for Odia character	[43]		Statistical features as euclidian distance, Hamilton distance	RBFNN	98.30
	[44]		LeNet-5 CNN	HopField NN	95.00
	Proposed work (without augmentation)		Features obtained from convolutional layer	CNN	97.76
	Proposed work (with augmentation)	Translation	Features obtained from convolutional layer	CNN	98.58
		Rotation			98.24
		Scaling			98.18
		Elastic deformation			98.89
		Noise			98.91
		Color inversion			98.58