Table 3 Comparison of object detection \(AP_{50}\) and inference time for the Mask R-CNN and YOLO models.
From: Multipurpose monitoring system for edible insect breeding based on machine learning
Model | \(AP_{50}\) | Inference time/tile | ||||
|---|---|---|---|---|---|---|
Live larvae | Pupae | Beetles | Dead larvae | Pests | ||
Mask R-CNN (ResNet-101) | 0.915 | 0.900 | 0.934 | 0.814 | 0.777 | 120 ms |
YOLOv5x | 0.894 | 0.893 | 0.921 | 0.700 | 0.738 | 40 ms |
