Table 6 Performance analysis for proposed method with existing based on database 1 (80% variation in training rate).

From: Attention-based hybrid deep learning model with CSFOA optimization and G-TverskyUNet3+ for Arabic sign language recognition

Approaches

Accuracy

Precision

Sensitivity

Specificity

MCC

NPV

F-measure

FPR

FNR

Proposed

0.99927

0.99967

0.98173

0.98928

0.98327

0.98051

0.98756

0.01827

0.00957

SVM30

0.96284

0.95283

0.95894

0.95889

0.95081

0.95827

0.95519

0.03349

0.03981

AArSLRS31

0.95842

0.94183

0.94583

0.95083

0.94193

0.95992

0.95083

0.04893

0.04015

ArSL-CNN32

0.96342

0.95083

0.96583

0.95284

0.96048

0.95183

0.95008

0.03156

0.02832

CNN35

0.94827

0.94883

0.95583

0.94482

0.95083

0.94144

0.950159

0.05892

0.04188

Back to article page