Multi-Scale Object Detection in Satellite Images Based on Improved YOLOv7

标签	类别	训练集规模/张	评估集规模/张	测试集规模/张	数据总量/张	目标平均尺度/像素
1	船舶	950	948	1 988	3 886	167.44
2	桥梁	1 123	1 121	2 326	4 570	246.10
3	田径场	859	856	2 017	3 734	276.50
4	储油罐	1 099	1 092	2 215	4 406	125.60
5	篮球场	923	920	2 033	3 876	108.19
6	网球场	1 043	1 040	2 212	4 295	102.71
7	飞机	1 226	1 222	2 451	4 899	113.21
8	棒球场	1 007	1 004	2 022	4 033	231.61
9	港口	967	964	1 953	3 884	163.96
10	车辆	1 188	1 186	2 382	4 756	41.96
11	十字路口	903	901	2 219	4 023	220.54
12	丁字路口	1 066	1 065	2 289	4 420	198.71
13	停车场	1 241	1 237	2 480	4 958	122.85

Table 2. Comparison of results on different models %
View table
View in Article
Table 2. Comparison of results on different models %
模型 YOLOv7 HRCNN Fast R-CNN+GACL-Net Faster R-CNN+GACL-Net YOLOv7+CNeB
飞机 96.0 82.9 83.3 90.8 98.4
船舶 68.7 57.6 75.0 88.5 86.5
储油罐 82.2 78.6 79.8 89.2 88.4
车辆 60.6 42.8 46.1 86.9 86.7
mAP 73.4 65.5 71.1 88.9 90.0

Tools

Get Citation

Copy Citation Text

Yuhao WEI, Song HUANG, Yani HUANG. Multi-Scale Object Detection in Satellite Images Based on Improved YOLOv7[J]. Spacecraft Recovery & Remote Sensing, 2024, 45(2): 153

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites