Back to Search
Start Over
BAFormer: A Novel Boundary-Aware Compensation UNet-like Transformer for High-Resolution Cropland Extraction.
- Source :
- Remote Sensing; Jul2024, Vol. 16 Issue 14, p2526, 27p
- Publication Year :
- 2024
-
Abstract
- Utilizing deep learning for semantic segmentation of cropland from remote sensing imagery has become a crucial technique in land surveys. Cropland is highly heterogeneous and fragmented, and existing methods often suffer from inaccurate boundary segmentation. This paper introduces a UNet-like boundary-aware compensation model (BAFormer). Cropland boundaries typically exhibit rapid transformations in pixel values and texture features, often appearing as high-frequency features in remote sensing images. To enhance the recognition of these high-frequency features as represented by cropland boundaries, the proposed BAFormer integrates a Feature Adaptive Mixer (FAM) and develops a Depthwise Large Kernel Multi-Layer Perceptron model (DWLK-MLP) to enrich the global and local cropland boundaries features separately. Specifically, FAM enhances the boundary-aware method by adaptively acquiring high-frequency features through convolution and self-attention advantages, while DWLK-MLP further supplements boundary position information using a large receptive field. The efficacy of BAFormer has been evaluated on datasets including Vaihingen, Potsdam, LoveDA, and Mapcup. It demonstrates high performance, achieving mIoU scores of 84.5%, 87.3%, 53.5%, and 83.1% on these datasets, respectively. Notably, BAFormer-T (lightweight model) surpasses other lightweight models on the Vaihingen dataset with scores of 91.3% F1 and 84.1% mIoU. [ABSTRACT FROM AUTHOR]
- Subjects :
- CONVOLUTIONAL neural networks
TRANSFORMER models
DEEP learning
REMOTE sensing
FARMS
Subjects
Details
- Language :
- English
- ISSN :
- 20724292
- Volume :
- 16
- Issue :
- 14
- Database :
- Complementary Index
- Journal :
- Remote Sensing
- Publication Type :
- Academic Journal
- Accession number :
- 178698053
- Full Text :
- https://doi.org/10.3390/rs16142526