Table 5 Performance evaluation of representative object detection architectures in wide-angle sensor sub-dataset.
From: MMFW-UAV dataset: multi-sensor and multi-view fixed-wing UAV dataset for air-to-air vision tasks
Methods | Recall | Precision | F1-Score | AP | FPS |
|---|---|---|---|---|---|
Yolo V4-Tiny | 98.44% | 98.75% | 0.99 | 99.82% | 418.69 |
Yolo V7-Tiny | 98.12% | 99.37% | 0.99 | 99.54% | 292.16 |
Faster-RCNN | 52.81% | 43.56% | 0.48 | 45.77% | 49.76 |
Yolo X | 97.50% | 99.05% | 0.98 | 99.25% | 248.80 |
Detr | 17.19% | 16.92% | 0.17 | 9.70% | 42.06 |
