Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation
The introduction of an attention mechanism in remote sensing image segmentation improves the accuracy of the segmentation. In this paper, a novel multi-scale KAN-based linear attention (MKLA) segmentation network of MKLANet is developed to promote a better segmentation result. A hybrid global–local...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-02-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/17/5/802 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850053246433886208 |
|---|---|
| author | Yuanhang Li Shuo Liu Jie Wu Weichao Sun Qingke Wen Yibiao Wu Xiujuan Qin Yanyou Qiao |
| author_facet | Yuanhang Li Shuo Liu Jie Wu Weichao Sun Qingke Wen Yibiao Wu Xiujuan Qin Yanyou Qiao |
| author_sort | Yuanhang Li |
| collection | DOAJ |
| description | The introduction of an attention mechanism in remote sensing image segmentation improves the accuracy of the segmentation. In this paper, a novel multi-scale KAN-based linear attention (MKLA) segmentation network of MKLANet is developed to promote a better segmentation result. A hybrid global–local attention mechanism in a feature decoder is designed to enhance the ability of aggregating the global–local context and avoiding potential blocking artifacts for feature extraction and segmentation. The local attention channel adopts MKLA block by bringing the merits of KAN convolution in Mamba like the linear attention block to improve the ability of handling linear and nonlinear feature and complex function approximation with a few extra computations. The global attention channel uses long-range cascade encoder–decoder block, where it mainly employs the 7 × 7 depth-wise convolution token mixer and lightweight 7 × 7 dilated deep convolution to capture the long-distance spatial features field and retain key spatial information. In addition, to enrich the input of the attention block, a deformable convolution module is developed between the encoder output and corresponding scale decoder, which can improve the expression ability of the segmentation model without increasing the depth of the network. The experimental results of the Vaihingen dataset (83.68% in mIoU, 92.98 in OA, and 91.08 in mF1), the UAVid dataset (69.78% in mIoU, 96.51 in OA), the LoveDA dataset (51.53% in mIoU, 86.42% in OA, and 67.19% in mF1), and the Potsdam dataset (97.14% in mIoU, 92.64% in OA, and 93.8% in mF1) outperform other advanced attention-based approaches in terms of small targets and edges’ segmentation. |
| format | Article |
| id | doaj-art-197e36bd9b7d4b038a8dcc60fce1745a |
| institution | DOAJ |
| issn | 2072-4292 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj-art-197e36bd9b7d4b038a8dcc60fce1745a2025-08-20T02:52:35ZengMDPI AGRemote Sensing2072-42922025-02-0117580210.3390/rs17050802Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic SegmentationYuanhang Li0Shuo Liu1Jie Wu2Weichao Sun3Qingke Wen4Yibiao Wu5Xiujuan Qin6Yanyou Qiao7National Engineering Research Center for Geomatics (NCG), Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, ChinaNational Engineering Research Center for Geomatics (NCG), Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, ChinaNational Engineering Research Center for Geomatics (NCG), Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, ChinaNational Engineering Research Center for Geomatics (NCG), Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, ChinaNational Engineering Research Center for Geomatics (NCG), Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, ChinaBeijing Institute of Control and Electronic Technology, Beijing 100038, ChinaBeijing Institute of Control and Electronic Technology, Beijing 100038, ChinaNational Engineering Research Center for Geomatics (NCG), Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, ChinaThe introduction of an attention mechanism in remote sensing image segmentation improves the accuracy of the segmentation. In this paper, a novel multi-scale KAN-based linear attention (MKLA) segmentation network of MKLANet is developed to promote a better segmentation result. A hybrid global–local attention mechanism in a feature decoder is designed to enhance the ability of aggregating the global–local context and avoiding potential blocking artifacts for feature extraction and segmentation. The local attention channel adopts MKLA block by bringing the merits of KAN convolution in Mamba like the linear attention block to improve the ability of handling linear and nonlinear feature and complex function approximation with a few extra computations. The global attention channel uses long-range cascade encoder–decoder block, where it mainly employs the 7 × 7 depth-wise convolution token mixer and lightweight 7 × 7 dilated deep convolution to capture the long-distance spatial features field and retain key spatial information. In addition, to enrich the input of the attention block, a deformable convolution module is developed between the encoder output and corresponding scale decoder, which can improve the expression ability of the segmentation model without increasing the depth of the network. The experimental results of the Vaihingen dataset (83.68% in mIoU, 92.98 in OA, and 91.08 in mF1), the UAVid dataset (69.78% in mIoU, 96.51 in OA), the LoveDA dataset (51.53% in mIoU, 86.42% in OA, and 67.19% in mF1), and the Potsdam dataset (97.14% in mIoU, 92.64% in OA, and 93.8% in mF1) outperform other advanced attention-based approaches in terms of small targets and edges’ segmentation.https://www.mdpi.com/2072-4292/17/5/802semantic segmentationremote sensing image segmentationattention mechanismdeformable convolutionmulti-scale feature fusion |
| spellingShingle | Yuanhang Li Shuo Liu Jie Wu Weichao Sun Qingke Wen Yibiao Wu Xiujuan Qin Yanyou Qiao Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation Remote Sensing semantic segmentation remote sensing image segmentation attention mechanism deformable convolution multi-scale feature fusion |
| title | Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation |
| title_full | Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation |
| title_fullStr | Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation |
| title_full_unstemmed | Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation |
| title_short | Multi-Scale Kolmogorov-Arnold Network (KAN)-Based Linear Attention Network: Multi-Scale Feature Fusion with KAN and Deformable Convolution for Urban Scene Image Semantic Segmentation |
| title_sort | multi scale kolmogorov arnold network kan based linear attention network multi scale feature fusion with kan and deformable convolution for urban scene image semantic segmentation |
| topic | semantic segmentation remote sensing image segmentation attention mechanism deformable convolution multi-scale feature fusion |
| url | https://www.mdpi.com/2072-4292/17/5/802 |
| work_keys_str_mv | AT yuanhangli multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT shuoliu multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT jiewu multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT weichaosun multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT qingkewen multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT yibiaowu multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT xiujuanqin multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation AT yanyouqiao multiscalekolmogorovarnoldnetworkkanbasedlinearattentionnetworkmultiscalefeaturefusionwithkananddeformableconvolutionforurbansceneimagesemanticsegmentation |