mlsanityData to 13 Oct 2026

GMBFormer: An NDVI-Guided Global Memory Bank Transformer for Urban Green-Space Extraction from Ultra-High-Resolution Imagery

arXiv 2606.06363 ↗First posted 4 Jun 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Urban green-space extraction from ultra-high-resolution (UHR) imagery is commonly performed patch by patch, which limits semantic reuse among spatially separated but visually similar vegetation patterns. Directly injecting the Normalized Difference Vegetation Index (NDVI) into red-green-blue (RGB) backbones can also blur the roles of visual appearance learning and physical vegetation confidence. We propose GMBFormer, a SegFormer-based framework that replaces adjacency-driven feature propagation with selective, similarity-driven prototype retrieval. Only RGB channels enter the backbone and decoder, while NDVI is decoupled as a physics-informed gate that admits high-confidence vegetation descriptors into a compact global memory bank through momentum updates. During training and inference, the current patch queries stored prototypes through memory-mediated cross-attention, and the retrieved response is integrated with bounded overhead. Experiments use a self-constructed Chengdu UHR dataset with 7,700 labeled 512 x 512 patches and two reduced-label settings derived from the public International Society for Photogrammetry and Remote Sensing (ISPRS) Potsdam dataset. Under the same training and evaluation protocol, GMBFormer obtains mean intersection over union (mIoU)/mean Dice (mDice) scores of 89.25%/94.31%, 92.17%/95.92%, and 83.72%/90.86%, respectively, improving the controlled SegFormer-B4 baseline in each setting. Ablation studies indicate that decoupled NDVI admission, memory retrieval, capacity, and momentum jointly shape the final performance.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 18 of its 37 references that are in the map. First posted on arXiv 4 Jun 2026.

Builds on

18 of 37 references.

References · 18

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,656
  2. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,111
  3. 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,849
  4. 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,658
  5. 2018Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentationcited by 3,216
  6. 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,930
  7. 2019PANet: Few-Shot Image Semantic Segmentation with Prototype Alignmentcited by 449
  8. 2019Video Object Segmentation using Space-Time Memory Networkscited by 437
  9. 2022XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Modelcited by 437
  10. 2019ResUNet-a: a deep learning framework for semantic segmentation of remotely sensed datacited by 261
  11. 2021Adaptive Prototype Learning and Allocation for Few-Shot Segmentationcited by 207
  12. 2021Rethinking Space-Time Networks with Improved Memory Coverage for Efficient Video Object Segmentationcited by 202
6 more
  1. 2021UNetFormer: A UNet-like Transformer for Efficient Semantic Segmentation of Remote Sensing Urban Scene Imagerycited by 166
  2. 2018Land-Cover Classification with High-Resolution Remote Sensing Images Using Transferable Deep Modelscited by 118
  3. 2021Beyond Self-attention: External Attention using Two Linear Layers for Visual Taskscited by 109
  4. 2019Collaborative Global-Local Networks for Memory-Efficient Segmentation of Ultra-High Resolution Imagescited by 51
  5. 2022Pin the Memory: Learning to Generalize Semantic Segmentationcited by 48
  6. 2021Attentional Prototype Inference for Few-Shot Segmentationcited by 6

Built on by

Not cited yet.