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

Table 5

Detection time.

	Detection time^1 (s)
	Faster R-CNN			Mask R-CNN
Diseases	VGG-16	ResNet-50	ResNet-101	MobileNet	ResNet-50	ResNet-101
Healthy tomato	0.158	0.160	0.174	0.088	0.166	0.186
Tomato malformed fruit	0.164	0.177	0.192	0.103	0.189	0.200
Tomato blotchy ripening	0.234	0.242	0.268	0.143	0.261	0.279
Tomato puffy fruit	0.209	0.226	0.243	0.164	0.236	0.253
Tomato dehiscent fruit	0.165	0.173	0.193	0.109	0.184	0.204
Tomato blossom-end rot	0.238	0.240	0.264	0.143	0.250	0.275
Tomato sunscald	0.233	0.262	0.291	0.161	0.282	0.311
Tomato virus disease	0.205	0.213	0.222	0.114	0.222	0.232
Tomato gray mold	0.218	0.227	0.231	0.123	0.238	0.242
Tomato ulcer disease	0.318	0.326	0.365	0.220	0.345	0.377
Tomato anthracnose	0.411	0.490	0.547	0.212	0.585	0.667
Mean time	0.202	0.209	0.226	0.123	0.227	0.246

¹The detection time of the model on each tomato disease types; bold face indicates the minimum detection time of the model.