Table 5 Comparative experiments of different models on WiderPerson dataset.
From: A small object detection model in aerial images based on CPDD-YOLOv8
Model | Precision (%) | Recall (%) | F1 (%) | mAP@0.5 (%) | mAP@0.5:0.95 (%) | Parameters (M) | FPS (f/s) | GFLOPs |
|---|---|---|---|---|---|---|---|---|
SSD | 86.3 | 34.3 | 49.1 | 51.5 | 21.2 | 24 | 7 | 24 |
Faster R-CNN | 54.3 | 64.5 | 59.0 | 61.5 | 30.1 | 250 | 32 | 250 |
YOLOv5 | 78.0 | 60.5 | 68.2 | 70.4 | 41.5 | 62 | 16 | 12 |
YOLOv7-tiny | 78.3 | 62.0 | 69.2 | 71.5 | 41.1 | 60 | 32 | 60 |
YOLOv8 | 79.6 | 62.1 | 69.8 | 73.2 | 45.4 | 111 | 52 | 111 |
YOLOv10 | 78.6 | 61.3 | 68.8 | 72.5 | 45.2 | 80.3 | 60 | 24.5 |
CPDD-YOLOv8 | 79.0 | 64.1 | 70.8 | 74.5 | 47.0 | 206 | 22 | 141.5 |
