cs.CVSep 19, 2026

SatOV: Restoring Spatial Priors for Training-Free Open-Vocabulary Segmentation in Remote Sensing Imagery

Authors: Changhao Zhao, Linglin Zeng, Hai Liu

Organizations: College of Resources and Environment, Huazhong Agricultural University, Wuhan 430070, China

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Open-vocabulary semantic segmentation (OVS) of remote sensing imagery is a challenging pixel-level task requiring strong generalization and adaptation to the spatial characteristics of remote sensing data. Although existing vision-language foundation models perform well in general domains, their image-level classification design weakens the spatial priors needed for high-resolution remote sensing segmentation: structural spatial relations are degraded during deep feature transformation, and fine-grained spatial details are lost during downsampling. To address these complementary deficiencies, we propose SatOV, a training-free framework for open-vocabulary remote sensing segmentation that restores spatial priors at two stages of the representation pipeline. Specifically, Residual QQ Attention (ResQQ) extracts Query-Key self-attention from an intermediate CLIP layer and fuses it with final-layer Query-Query attention via a residual combination, restoring structural spatial priors suppressed by the final-layer representation. Spatially Modulated Upsampling (SatUp) uses the original high-resolution RGB image as spatial guidance, combining spatial feature modulation with guided cross-attention to reconstruct pixel-level textures and boundaries. Extensive experiments on DOTA, UDD, LoveDA, and Vaihingen show that SatOV consistently improves training-free OVS and achieves competitive quantitative and qualitative results against state-of-the-art methods. These results validate the effectiveness of restoring spatial priors at both the representation and spatial-resolution stages for remote sensing open-vocabulary segmentation.

Figures & tables

Figure 1: Experimental analysis of spatial covariance across Transformer depth, showing the progressive loss of spatial detail with increasing semantic abstraction.
Figure 2: Left panel illustrates the restoration of structural priors within the CLIP ViT: ResQQ extracts the Query‑Key attention from the penultimate layer as the overwritten spatial prior, and fuses it with the final‑layer Query‑Query self‑attention via residual connections, recovering the spatial covariance of the low‑resolution features. Right panel demonstrates the recovery of pixel‑level detail priors: SatUp uses the original high‑resolution RGB image as a spatial index, and applies Spatial Feature Modulation (SFT) together with cross‑attention to upsample the low‑resolution semantic features back to pixel‑level space.
Figure 3: SatUp takes low-resolution CLIP features and the original high-resolution RGB image as inputs. It encodes the image into spatial queries, applies rotary positional encoding, and performs cross-attention with SFT-modulated keys and semantic values, producing pixel-aligned high-resolution features.
DatasetImaging PlatformPrimary Type
UDDUAVUrban
DOTASatelliteDense objects
LoveDASatelliteFarmland and Urban
VaihingenAerial (no blue band)Building and House
Table 1: Overview of evaluation datasets.
ModelUDDDOTALoveDAVaihingenAvg.
Training-Free Methods
MaskCLIP ECCV baseline32.425.127.823.427.2
SCLIP ECCV38.729.330.429.131.9
ClearCLIP ECCV41.231.031.635.534.8
LPOSS CVPR41.833.932.438.036.5
SegEarth-OV CVPR45.340.336.942.341.2
Table 2: Performance comparison of different methods on four remote sensing datasets. The primary comparison is conducted among training-free methods under the same zero-shot evaluation setting.
Figure 4: For the segmentation mask comparison on the DOTA dataset with densely packed objects, it can be observed that MaskCLIP predictions are largely semantically correct but completely lack spatial structure. ClearCLIP shows some improvement yet still suffers from internal holes. In contrast, SatOV accurately delineates the boundaries of densely arranged buildings and narrow roads, demonstrating the effectiveness of ResQQ and SatUp in jointly recovering spatial priors.
Figure 5: These are binary classification attention maps. The second row shows ClearCLIP outputs while the third row shows our SatOV. Significant differences arise from the resolution bottleneck, and SatOV achieves true pixel-level fine-grained prediction.
Figure 6: Visualization of feature reconstruction across different upsampling methods. It can be observed that SatUp achieves the best fine-grained reconstruction.
ModelPre-training DataUDD
+DINO CVPRImageNet41.8
+FeatUp ICLRImageNet45.3
+AnyUp ICLRImageNet44.2
+SatUp OursMillion-AID45.9
Table 3: Performance comparison of different upsamplers on the UDD dataset. Taking DINO as the baseline, all methods yield performance improvements, yet SatUp achieves the best results for remote sensing tasks.
ConfigurationDOTA∇\nabla
Full SatOV40.4—
−- SatUp34.8∇5.6\nabla 5.6
−- ResQQ32.8∇2.0\nabla 2.0
−- CLS token debiasing30.9∇1.9\nabla 1.9
−- FFN and Residual Removing30.0∇0.9\nabla 0.9
−- Query–Query attention29.0∇1.0\nabla 1.0
Table 4: Ablation study on the proposed components. Starting from the full SatOV model, we progressively remove each component to quantify its individual contribution. All models are evaluated under the same training-free, zero-shot setting on the DOTA dataset.
Figure 7: Visualization results for different ablation configurations. It can be seen that ResQQ can recover CLIP features to a certain extent, while SatUp leverages guidance from the original image to achieve full pixel-level granularity.
More in this topic

Open-Vocabulary Semantic Segmentation

Open-vocabulary semantic segmentation labels image pixels using text-defined categories beyond a fixed training vocabulary.

See all 84 papers

Explore similar work

CardsList

  1. Standalone DINOv3 for Training-Free Open-Vocabulary Semantic Segmentation in Remote Sensing

    Aug 4, 2026Changhao Zhao, Haoxiang Li, Yuke Li +2Remote Sensing Image SegmentationOpen-Vocabulary Semantic Segmentation

    College of Resources and Environment, Huazhong Agricultural University, Wuhan 430070, China

  2. ConInfer: Context-Aware Inference for Training-Free Open-Vocabulary Remote Sensing Segmentation

    Mar 31, 2026Wenyang Chen, Zhanxuan Hu, Yaping Zhang +2Image SegmentationRemote Sensing Image Segmentation

    Yunnan Normal University · Xi’an University of Posts & Telecommunications

  3. Open-Vocabulary Semantic Segmentation in Remote Sensing via Hierarchical Attention Masking and Model Composition

    Feb 27, 2026Mohammadreza Heidarianbaei, Mareike Dorozynski, Hubert Kanyamahanga +2Open-Vocabulary Semantic SegmentationRemote Sensing Image Segmentation

    Institute of Photogrammetry and GeoInformation Leibniz University Hannover Germany