基于多尺度特征增强的航拍小目标检测算法
|
|
肖剑,何昕泽,程鸿亮,杨小苑,胡欣
|
Aerial small target detection algorithm based on multi-scale feature enhancement
|
|
Jian XIAO,Xinze HE,Hongliang CHENG,Xiaoyuan YANG,Xin HU
|
|
| 表 5 不同算法在VisDrone数据集上的平均精度和参数量对比 |
| Tab.5 Comparative results of different algorithms in average precision and parameters on VisDrone 2019 dataset |
|
| 模型 | AP/% | mAP0.5/% | F/(帧$ \cdot {{\mathrm{s}}^{ - 1}}$) | Para/106 | | 行人 | 人 | 自行车 | 汽车 | 面包车 | 卡车 | 三轮车 | 遮阳棚三轮车 | 巴士 | 摩托车 | | Faster R-CNN[20] | 20.9 | 14.8 | 7.3 | 51.0 | 29.7 | 19.5 | 14.0 | 8.8 | 30.5 | 21.2 | 21.8 | 14.4 | — | | YOLOv5s | 39.0 | 31.3 | 11.2 | 73.5 | 35.4 | 29.5 | 20.5 | 11.1 | 43.1 | 37.0 | 33.2 | 118.0 | 7.03 | | TPH-YOLOv5[1] | 53.3 | 42.1 | 21.1 | 83.7 | 45.2 | 42.5 | 33.0 | 16.3 | 61.1 | 51.0 | 44.9 | 34.0 | 60.42 | | YOLOv7-tiny[21] | 37.9 | 34.6 | 9.4 | 76.1 | 36.3 | 29.8 | 20.1 | 10.6 | 43.2 | 41.8 | 34.0 | 89.0 | 6.03 | | YOLOv8s | 42.0 | 32.8 | 12.5 | 79.4 | 44.7 | 35.5 | 26.9 | 17.1 | 54.0 | 43.3 | 38.8 | 17.3 | 11.13 | | YOLOv8l | 51.1 | 39.8 | 21.9 | 82.9 | 49.3 | 45.4 | 38.1 | 20.4 | 67.0 | 52.3 | 46.8 | 59.0 | 43.69 | | YOLOv9-C[22] | 34.0 | 18.4 | 15.4 | 77.5 | 45.2 | 54.1 | 24.8 | 24.1 | 64.9 | 38.3 | 39.7 | — | 50.90 | | YOLOv11s | 41.6 | 31.8 | 11.2 | 79.5 | 45.4 | 35.5 | 26.1 | 15.5 | 55.1 | 43.3 | 38.5 | 121.8 | 9.46 | | 本研究模型 | 53.9 | 48.3 | 27.5 | 81.3 | 53.6 | 49.3 | 40.0 | 24.5 | 71.9 | 52.8 | 50.3 | 57.3 | 8.08 |
|
|
|