Table 2 Results of the detection growth stages of Tenebrio molitor (live larvae, pupae, beetles) and anomalies (dead larvae, pests) for the Mask R-CNN with the ResNet-101 backbone.
From: Multipurpose monitoring system for edible insect breeding based on machine learning
Type | \(AP_{50}\) | \(F1_{obj}\) | \(PPV_{obj}\) | \(TPR_{obj}\) | |
|---|---|---|---|---|---|
Live larvae (Tenebrio molitor) | Growth-stage | 0.915 | 0.905 | 0.927 | 0.884 |
Pupae (Tenebrio molitor) | 0.900 | 0.893 | 0.929 | 0.859 | |
Beetles (Tenebrio molitor) | 0.934 | 0.930 | 0.984 | 0.882 | |
Dead larvae (Tenebrio molitor) | Anomaly | 0.814 | 0.858 | 0.898 | 0.821 |
Pests (Alphitobius diaperinus) | 0.777 | 0.835 | 0.889 | 0.787 |
