Table 6 Performances of different YOLOv8 versions in this work.
From: An optimized lightweight real-time detection network model for IoT embedded devices
Version | Precision (%) | Recall (%) | mAP (%) | Params (M) | FLOPs (G) |
|---|---|---|---|---|---|
FRYOLO based on YOLOv8n | 84.7 | 89.0 | 92.5 | 3.2 | 8.7 |
FRYOLO based on YOLOv8s | 89.4 | 87.5 | 94.7 | 11.2 | 28.6 |
FRYOLO based on YOLOv8m | 89.8 | 88.3 | 94.8 | 25.9 | 78.9 |
FRYOLO based on YOLOv8l | 84.7 | 86.0 | 93.0 | 43.7 | 165.2 |
FRYOLO based on YOLOv8x | 87.6 | 86.8 | 93.8 | 68.2 | 257.8 |
