Table 3 Segmentation results of the different backbones
From: Attention-optimized DeepLab V3 + for automatic estimation of cucumber disease severity
Backbone networks | Class | Precision (%) | Recall (%) | F-1 (%) | IoU (%) |
|---|---|---|---|---|---|
ResNet101 | Background | 95.63 | 95.1 | 95.36 | 91.14 |
Leaf | 97.27 | 96.62 | 96.94 | 94.07 | |
Lesion | 65.72 | 84.32 | 73.87 | 58.56 | |
Xception | Background | 90.97 | 76.54 | 83.13 | 71.13 |
Leaf | 89.87 | 93.79 | 91.79 | 84.82 | |
Lesion | 43.61 | 83.82 | 57.37 | 40.23 | |
MobilenetV2 | Background | 95.36 | 91.1 | 93.18 | 87.23 |
Leaf | 95.69 | 96.57 | 96.13 | 92.55 | |
Lesion | 62.45 | 84.62 | 71.86 | 56.08 | |
MobilenetV3 | Background | 92.28 | 89.16 | 90.70 | 82.98 |
Leaf | 94.46 | 94.68 | 94.57 | 89.69 | |
Lesion | 55.74 | 78.55 | 65.21 | 48.37 | |
SE-ResNet50 | Background | 89.52 | 89.00 | 89.26 | 80.6 |
Leaf | 94.38 | 93.98 | 94.18 | 89.01 | |
Lesion | 62.82 | 76.14 | 68.84 | 52.48 | |
CBAM-ResNet50 | Background | 95.72 | 90.49 | 93.03 | 86.97 |
Leaf | 95.18 | 97.00 | 96.08 | 92.45 | |
Lesion | 68.72 | 82.1 | 74.82 | 59.76 | |
Proposed | Background | 94.71 | 93.66 | 94.18 | 89.01 |
Leaf | 97.01 | 96.96 | 96.98 | 94.14 | |
Lesion | 71.03 | 80.41 | 75.43 | 60.55 |