Full
Outline
To address the challenges of complex surface textures, blurred boundaries, and significant illumination variations in unstructured road scenarios, this paper proposes a freespace detection method using only red-green-blue (RGB) images and integrating boundary enhancement and geometry-depth prior attention. Based on the SegFormer framework, the proposed method extracts explicit boundary information through auxiliary boundary supervision and introduces boundary cues into the semantic feature space via a boundary feature fusion module, thereby enhancing the representation of soft boundaries and irregular contours. Meanwhile, a geometry-depth prior attention module is designed to incorporate geometric continuity priors and vertical pseudo-depth priors, enabling structured feature reweighting and improving prediction stability in complex terrain. Experimental results show that the proposed method achieves an IoU of 90.8% and an F1-score of 95.1% on the Off-Road Freespace Detection (ORFD) dataset, outperforming the compared monocular RGB methods on major segmentation metrics and demonstrating good cross-dataset generalization capability. These results indicate that the proposed method achieves a favorable balance among segmentation accuracy, boundary quality, and inference efficiency.
168
37
| 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
010-62662699(Operation and Technical Support)
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
