Table 7 Comparative experimental results of YOLO series models on the tinyperson dataset.
From: LPAE-YOLOv8: lightweight aerial small object detection via LSE-Head and adaptive attention
Models | P | R | mAP(0.5)(%) | mAP(0.5:0.95)(%) | The number of model parameters (×106) | GFLOPs (×109) |
|---|---|---|---|---|---|---|
YOLOv3 | 31.1% | 8.0% | 8.2% | 3.2% | 12.1 | 19.1 |
YOLOv5 | 34.2% | 16.7% | 14.0% | 5.0% | 2.51 | 7.2 |
YOLOv6 | 31.4% | 14.7% | 11.8% | 4.1% | 4.24 | 11.9 |
YOLOv8 | 35.4% | 16.8% | 14.4% | 5.0% | 3.01 | 8.1 |
YOLOv10 | 30.8% | 15.4% | 12.1% | 4.2% | 2.71 | 8.4 |
LPAE-YOLOv8 | 36.3% | 22.1% | 18.7% | 6.5% | 1.97 | 10.5 |
