Table 7 Comparative Analysis of various pre-trained CNN models with mean quantitative metrics of feature heatmap image with most significant 10, 12, 14 features of LIME.

From: Evaluation of deep learning models using explainable AI with qualitative and quantitative analysis for rice leaf disease detection

Number of features	Pre-trained models	Quantitative Metrics
Number of features	Pre-trained models	IoU	DSC	Sensitivity	Specificity	MCC	PWA	MAE
10	ResNet50	0.432	0.594	0.507	0.931	0.491	0.801	0.198
	InceptionResNetV2	0.421	0.586	0.498	0.928	0.482	0.793	0.206
	VGG16	0.392	0.551	0.477	0.907	0.424	0.778	0.221
	Xception	0.397	0.561	0.479	0.913	0.439	0.776	0.223
	DenseNet201	0.364	0.525	0.456	0.898	0.389	0.751	0.249
	AlexNet	0.365	0.512	0.444	0.895	0.371	0.758	0.242
	EfficientNetB0	0.326	0.474	0.418	0.874	0.315	0.729	0.271
	InceptionV3	0.295	0.441	0.371	0.878	0.284	0.723	0.276
12	ResNet50	0.467	0.611	0.561	0.906	0.491	0.799	0.201
	InceptionResNetV2	0.456	0.621	0.556	0.909	0.489	0.798	0.201
	VGG16	0.398	0.556	0.527	0.873	0.405	0.773	0.227
	Xception	0.412	0.576	0.516	0.886	0.437	0.783	0.201
	DenseNet201	0.361	0.521	0.483	0.861	0.358	0.741	0.258
	AlexNet	0.374	0.526	0.491	0.862	0.363	0.751	0.249
	EfficientNetB0	0.351	0.506	0.472	0.854	0.335	0.731	0.268
	InceptionV3	0.308	0.454	0.407	0.850	0.276	0.714	0.285
14	ResNet50	0.471	0.613	0.604	0.876	0.475	0.799	0.201
	InceptionResNetV2	0.466	0.630	0.591	0.872	0.477	0.792	0.207
	VGG16	0.398	0.557	0.545	0.845	0.391	0.771	0.228
	Xception	0.427	0.593	0.573	0.864	0.444	0.781	0.219
	DenseNet201	0.361	0.523	0.514	0.831	0.342	0.733	0.266
	AlexNet	0.384	0.537	0.529	0.834	0.359	0.747	0.252
	EfficientNetB0	0.358	0.517	0.51	0.825	0.331	0.722	0.277
	InceptionV3	0.31	0.461	0.439	0.818	0.263	0.700	0.299
Mean	ResNet50	0.457	0.606	0.554	0.904	0.486	0.800	0.200
	InceptionResNetV2	0.448	0.612	0.545	0.912	0.491	0.790	0.210
	VGG16	0.398	0.551	0.529	0.907	0.432	0.772	0.228
	Xception	0.393	0.553	0.491	0.896	0.419	0.777	0.223
	DenseNet201	0.352	0.512	0.465	0.891	0.389	0.758	0.242
	AlexNet	0.343	0.512	0.444	0.892	0.395	0.761	0.239
	EfficientNetB0	0.312	0.482	0.418	0.874	0.352	0.734	0.266
	InceptionV3	0.276	0.436	0.391	0.878	0.284	0.723	0.276

Back to article page

Table 7 Comparative Analysis of various pre-trained CNN models with mean quantitative metrics of feature heatmap image with most significant 10, 12, 14 features of LIME.

Search

Quick links