Table 3 The performances of the herbicide weed control spectrum neural networks for detecting and discriminating the sub-images containing weeds susceptible to ACCase-inhibiting herbicides, weeds susceptible to synthetic auxin herbicides, or bermudagrass turf exclusively (no herbicide)
From: Deep learning for detecting herbicide weed control spectrum in turfgrass
Deep learning architecture | Herbicides | Validation dataset | Testing dataset | ||||||
|---|---|---|---|---|---|---|---|---|---|
Precision | Recall | Overall accuracy | F1 score | Precision | Recall | Overall accuracy | F1 score | ||
GoogLeNet | ACCase-inhibiting | 0.995 | 0.999 | 0.998 | 0.997 | 0.993 | 0.999 | 0.997 | 0.996 |
Synthetic auxin | 0.999 | 0.995 | 0.998 | 0.997 | 0.998 | 0.994 | 0.997 | 0.996 | |
No herbicide | 1.000 | 0.999 | 1.000 | 0.999 | 1.000 | 0.999 | 1.000 | 0.999 | |
MobileNet-v3 | ACCase-inhibiting | 0.976 | 0.965 | 0.980 | 0.970 | 0.973 | 0.963 | 0.979 | 0.968 |
Synthetic auxin | 0.978 | 0.978 | 0.985 | 0.978 | 0.981 | 0.971 | 0.984 | 0.976 | |
No herbicide | 0.971 | 0.983 | 0.985 | 0.977 | 0.965 | 0.985 | 0.983 | 0.975 | |
ShuffleNet-v2 | ACCase-inhibiting | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 0.999 | 1.000 | 0.999 |
Synthetic auxin | 0.999 | 1.000 | 1.000 | 0.999 | 0.999 | 1.000 | 0.999 | 0.999 | |
No herbicide | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | |
VGGNet | ACCase-inhibiting | 0.998 | 1.000 | 0.999 | 0.999 | 0.998 | 0.999 | 0.999 | 0.998 |
Synthetic auxin | 1.000 | 1.000 | 1.000 | 1.000 | 0.998 | 1.000 | 0.999 | 0.999 | |
No herbicide | 1.000 | 0.998 | 0.999 | 0.999 | 1.000 | 0.997 | 0.999 | 0.998 | |