Research map / Detection and segmentation / CNN features
Revisiting Near/Remote Sensing With Geospatial Attention
Scott Workman, M. Usman Rafique, Hunter Blanton, Nathan Jacobs
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| combine overhead imagery and geospatially attended nearby panoramas | near/remote segmentation · Brooklyn land use | mIOU: 48.15% -> 69.24% | Improved |
| combine overhead imagery and geospatially attended nearby panoramas | near/remote segmentation · Brooklyn land cover | mIOU: 69.48% -> 74.59% | Improved |
| combine overhead imagery and geospatially attended nearby panoramas | near/remote segmentation · Queens land cover | mIOU: 62.63% -> 61.24% | Hurt |
| use geospatial attention and joint modalities | near/remote segmentation · Queens building function | mIOU: 3.73% -> 4.08% | Improved |
| replace PixelNet with LinkNet34 (ablation) | Brooklyn land use baseline · Brooklyn land use | mIOU: 45.54% -> 47.14% | Improved |
| replace VGG-16 with ResNet-50 (ablation) | LinkNet34 baseline · Brooklyn land use | mIOU: 47.14% -> 51.59% | Improved |
| replace LinkNet34 with EfficientUNet (ablation) | ours decoder · Brooklyn land use | mIOU: 67.43% -> 69.24% | Improved |
| add distance and orientation geometry to panorama-only attention (ablation) | geospatial attention · Brooklyn land use | mIOU: 53.77% -> 68.72% | Improved |
| add distance and orientation geometry to overhead-only attention (ablation) | geospatial attention · Brooklyn land use | mIOU: 53.66% -> 68.87% | Improved |
| add both distance and orientation rather than distance alone (ablation) | geospatial attention · Brooklyn land use | mIOU: 59.41% -> 68.46% | Improved |
| add both distance and orientation rather than orientation alone (ablation) | geospatial attention · Brooklyn land use | mIOU: 62.04% -> 68.46% | Improved |
| add panorama and overhead appearance features to geometry-only attention (ablation) | geospatial attention · Brooklyn land use | mIOU: 68.46% -> 69.24% | Improved |
Closest papers by meaning
- ReAttnCLIP: Training-Free Open-Vocabulary Remote Sensing Image Segmentation via Re-defined Attention in CLIP (CVPR 2026 · similarity 0.89)
- GeoViS: Geospatially Rewarded Visual Search for Remote Sensing Visual Grounding (CVPR 2026 · similarity 0.88)
- A Unified Model for Near and Remote Sensing (ICCV 2017 · similarity 0.88)
- Augmenting Depth Estimation With Geospatial Context (ICCV 2021 · similarity 0.86)
- See What We Cannot See: A Geo-guided Reasoning Benchmark for Object Counting under Adverse Earth Observation Conditions (CVPR 2026 · similarity 0.84)
- ZoomEarth: Active Perception for Ultra-High-Resolution Geospatial Vision-Language Tasks (CVPR 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
