Identification of Tomato Disease Types and Detection of Infected Areas Based on Deep Convolutional Neural Networks and Object Detection Techniques

Table 3

The and the values of the tomato images obtained by different detection architectures.

	Faster R-CNN (%)			Mask R-CNN (%)
Diseases	VGG-16	ResNet-50	ResNet-101	MobileNet	ResNet-50	ResNet-101
Healthy tomato	90.62	90.66	90.54	90.39	100.00	100.00
Tomato malformed fruit	94.20	99.82	100.00	100.00	100.00	100.00
Tomato blotchy ripening	100.00	100.00	100.00	100.00	100.00	100.00
Tomato puffy fruit	70.59	71.30	73.77	71.70	100.00	100.00
Tomato dehiscent fruit	100.00	100.00	100.00	100.00	98.88	100.00
Tomato blossom-end rot	70.00	97.80	98.20	97.50	100.00	100.00
Tomato sunscald	94.18	89.05	98.33	100.00	100.00	100.00
Tomato virus disease	77.96	75.89	73.45	77.44	99.52	100.00
Tomato gray mold	90.43	100.00	100.00	100.00	93.33	100.00
Tomato ulcer disease	79.80	67.17	83.47	67.00	100.00	100.00
Tomato anthracnose	79.24	80.86	53.10	68.26	92.00	96.00
	86.09	88.41	88.53	88.39	98.52	99.64

Bold faces are the detection results of the architecture with the best performance.