Performance analysis of dog breed prediction using Xception model over Densenet model

This research aims to conduct a performance analysis of Xception compared to DenseNet in predicting dog breeds with enhanced accuracy. Two groups, Xception and DenseNet, each comprising 20 samples, were utilized for implementation. Sample size determination was accomplished using Clincalc software with a GPower pretest of 80%, a threshold of 0.05, and a confidence interval of 95%. Results indicate that the accuracy of the Xception model is 91%, while that of DenseNet is 87%. The difference in significance between groups is 0.04 (Independent t-test, p<0.05), signifying statistical significance. Consequently, the Xception model achieves an accuracy of 91%, surpassing that of the DenseNet model, which stands at 87%.