PDF
Full
Outline
In remote sensing images of complex scenes, ships exhibit significant scale variations. In particular, their key regions are represented by only a few pixels, making direct detection methods susceptible to background noise interference, which results in insufficient accuracy and robustness. To address these challenges, a hierarchical detection method based on a Multi-level Detection Network (MDNet) is proposed. In the first stage, which is built upon Cascade R-CNN, a global context module is integrated to enhance scene discrimination capability. Furthermore, deformable convolutional heads are employed to adapt to the geometric variations of objects, through which precise coarse localization of ships is achieved. Following automated cropping and enhancement via Gamma Correction, a dual attention mechanism is utilized in the second stage to focus on the weak features within local image patches, whereby fine-grained identification of the key regions is performed. Through this method, complex background noise can be effectively filtered, and salient features in key regions can be focused on. A significant improvement in average precision is thus achieved compared to direct detection methods.
PDF
311
137
| 科 Family | 属数 Number of genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) | 属 Genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) |
|---|---|---|---|---|---|---|
| 鹅膏菌科Amanitaceae | 2 | 11 | 5.26 | 鹅膏菌属 Amanita | 10 | 4.78 |
| 小菇科 Mycenaceae | 2 | 12 | 5.74 | 丝盖伞属 Inocybe | 5 | 2.39 |
| 多孔菌科 Polyporaceae | 8 | 14 | 6.70 | 蜡蘑属 Laccaria | 5 | 2.39 |
| 红菇科 Russulaceae | 3 | 23 | 11.00 | 小皮伞属 Marasmius | 6 | 2.87 |
| 小菇属 Mycena | 11 | 5.26 | ||||
| 光柄菇属 Pluteus | 5 | 2.39 | ||||
| 红菇属 Russula | 17 | 8.13 | ||||
| 栓菌属 Trametes | 5 | 2.39 |
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62199257
qkjq@cast.org.cn
Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House
