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

Table 1

Details of the four deep convolutional neural networks.

DCNN	Parameters (M)	Number of layers	Top-1 error (%)^1	Top-5 error (%)^2
VGG-16	138	16	28.07	9.33
ResNet-50	25	50	22.85	6.71
ResNet-101	42.6	101	21.75	6.05
MobileNet	3.3	28	—	—

¹The Top-1 error rate on ImageNet Validation; ²the Top-5 error rate on ImageNet Validation; the symbol “-” indicates that no corresponding data are given in the original paper.