Improved CNN Method for Crop Pest Identification Based on Transfer Learning

Table 1

VGG 16 network structure model parameters.

Layer	Size	Filter size	Stride	Total parameters
Input layer	224 × 224 × 6	—	—	0
Conv1	224 × 224 × 128	3 × 3 × 3	1	1918
Conv2	224 × 224 × 128	3 × 3 × 64	1	45818
Maxpooling1	112 × 112 × 64	3×3	2	0
Conv3	112 × 112 × 256	3 × 3 × 128	1	732224
Conv4	112 × 112 × 256	3 × 3 × 256	1	148648
Maxpooling2	56 × 56 × 128	2 × 2	2	0
Conv5	56 × 56 × 128	3 × 3 × 64	1	28468
Conv6	56 × 56 × 128	3 × 3 × 128	1	59878
Conv7	56 × 56 × 128	3 × 3 × 128	1	59874
Maxpooling3	28 × 28 × 512	3 × 3 × 3	2	0
Conv8	28 × 28 × 512	3 × 3 × 3	1	1178956
Conv9	28 × 28 × 512	2 × 2	1	2485896
Conv10	28 × 28 × 512	3 × 3 × 512	1	2487678
Maxpooling4	14 × 14 × 256	2 × 2	2	0
Conv11	14 × 14 × 256	3 × 3 × 64	1	2359296
Conv12	14 × 14 × 256	3 × 3 × 128	1	2359296
Conv13	14 × 14 × 256	3 × 3 × 128	1	2359296
Maxpooling5	7 × 7 × 128	3 × 3 × 128	2	0
FC1	4096	—	—	102760448
FC2	4096	—	—	16777216
FC3	1024	—	—	4096000