Research map / Detection and segmentation / CNN features
TerrainGraphNet: Terrain-Constrained Graph Reasoning for Landslide Segmentation
SHAHID SHAFI DAR, Pranjal Pandey, Nagendra Kumar
Read the paperCheck 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 |
|---|---|---|---|
| add local multi-scale encoder to SAM backbone (ablation) | TerrainGraphNet · Bijie | IoU: 69.92 -> 70.20 | Improved |
| add terrain encoder with terrain-modulated feature interaction (ablation) | TerrainGraphNet · Bijie | IoU: 70.20 -> 72.04 | Improved |
| add hierarchical decoder (ablation) | TerrainGraphNet · Bijie | IoU: 72.04 -> 74.80 | Improved |
| add terrain-aware graph reasoning (TAGR) (ablation) | TerrainGraphNet · Bijie | IoU: 74.80 -> 79.08 | Improved |
| add TAGR to concat fusion (ablation) | TerrainGraphNet · Bijie | HD95: 18.20 -> 14.42 | Improved |
| use TMFI instead of element-wise addition (ablation) | TerrainGraphNet · Bijie | IoU: 77.32 -> 79.08 | Improved |
| use TMFI instead of element-wise multiplication (ablation) | TerrainGraphNet · Bijie | IoU: 76.21 -> 79.08 | Improved |
| use balanced alpha 0.5 instead of feature-only alpha 1.00 (ablation) | TerrainGraphNet · Bijie | IoU: 78.02 -> 79.08 | Improved |
| use slope in affinity (alpha 0.5) instead of slope-only alpha 0 (ablation) | TerrainGraphNet · Bijie | IoU: 77.55 -> 79.08 | Improved |
| add aspect or curvature descriptors beyond slope (ablation) | TerrainGraphNet · Bijie | IoU: 78.69 (aspect) vs 79.08 (slope) | No clear effect |
| increase graph neighbors and layers to K=16, L=3 (ablation) | TerrainGraphNet · Bijie | IoU: 79.08 -> 77.16 | Hurt |
| reduce graph layers to L=1 (ablation) | TerrainGraphNet · Bijie | IoU: 79.08 -> 78.83 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SegFormer — SegFormer: A Topic Segmentation Model with Controllable Range of Attention (AAAI 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Patched Line Segment Learning for Vector Road Mapping (AAAI 2024 · similarity 0.82)
- Terrain Diffusion Network: Climatic-Aware Terrain Generation with Geological Sketch Guidance (AAAI 2024 · similarity 0.82)
- Towards Faithful Reasoning in Remote Sensing: A Perceptually-Grounded GeoSpatial Chain-of-Thought for Vision-Language Models (ICLR 2026 · similarity 0.82)
- TerraScope: Pixel-Grounded Visual Reasoning for Earth Observation (CVPR 2026 · similarity 0.81)
- LithoBench: Benchmarking Large Multimodal Models for Remote-Sensing Lithology Interpretation (NeurIPS 2026 · similarity 0.81)
- SatSynth: Augmenting Image-Mask Pairs through Diffusion Models for Aerial Semantic Segmentation (CVPR 2024 · similarity 0.81)
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.
