Table 3 mAP metric score for different models developed from this study
Architecture | Model | Training time (h) | Accuracy |
|---|---|---|---|
Faster R-CNN InceptionV2 | Entire plant | 30 | 0.728579 |
Faster R-CNN InceptionV2 | Leaves | 30 | 0.701833 |
Faster R-CNN InceptionV2 | Pseudostem | 30 | 0.999447 |
Faster R-CNN InceptionV2 | Fruit bunch | 30 | 0.973025 |
Faster R-CNN InceptionV2 | Cut fruit | 30 | 0.953296 |
Faster R-CNN InceptionV2 | Corm | 30 | 0.979151 |
Faster R-CNN ResNet50 | Entire plant | 20 | 0.734611 |
Faster R-CNN ResNet50 | Leaves | 20 | 0.703871 |
Faster R-CNN ResNet50 | Pseudostem | 20 | 0.999905 |
Faster R-CNN ResNet50 | Fruit bunch | 20 | 0.973634 |
Faster R-CNN ResNet50 | Cut fruit | 20 | 0.941152 |
Faster R-CNN ResNet50 | Corm | 20 | 0.976888 |
SSD MobileNetV1 | Entire plant | 50 | 0.446880 |
SSD MobileNetV1 | Leaves | 50 | 0.619923 |
SSD MobileNetV1 | Pseudostem | 50 | 0.98239 |
SSD MobileNetV1 | Fruit bunch | 50 | 0.936463 |
SSD MobileNetV1 | Cut fruit | 50 | 0.923548 |
SSD MobileNetV1 | Corm | 50 | 0.997296 |