Tamil OCR Conversion from Digital Writing Pad Recognition Accuracy Improves through Modified Deep Learning Architectures

Table 7

Evaluation of the classifier’s performance through statistical measures with different age groups based on handwritten alphabet recognition.

Age group	Accuracy		Precision		Recall		F1 Score
	MMU-SNet (%)	RTSBA (%)	MMU-SNet	RTSBA (%)	MMU-SNet (%)	RTSBA (%)	MMU-SNet (%)	RTSBA (%)
Age 15–25	95.5	98.08	98.6%	99.35	96.7	98.7	97.6	99.02
Age 26–35	97.4	99.00	99.3%	99.04	98.0	98.7	98.7	99.35
Age 36–45	96.1	99.24	99.3	99.04	96.7	98.8	98.0	99.44
Age 46–55	96.8	99.36	98.6%	98.7	98.0	99.00	98.3	99.68
Age 56–65	94.2	97.44	97.3%	99.34	96.6	98.05	96.9	98.7

RTSBA, ResNet two-stage bottleneck architecture; MMU-SNet, modified multi-scale segmentation network.