Research map / Detection and segmentation / Open-set and few-shot detection
Generalizable Slum Detection from Satellite Imagery with Mixture-of-Experts
Sumin Lee, Sungwon Park, Jeasurk Yang, Jihee Kim, Meeyoung Cha
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 MoE layers to the source segmentation model | SegFormer source model · Dar es Salaam | mIoU: 0.681 -> 0.806 | Improved |
| add cross-expert test-time self-training on top of MoE source | MoE Source · Dar es Salaam | mIoU: 0.806 -> 0.859 | Improved |
| add mutual information regularization between region and expert (ablation) | GRAM · Dar es Salaam | mIoU: 0.734 -> 0.859 | Improved |
| add region classification loss Ldom (ablation) | GRAM · Dar es Salaam | mIoU: 0.836 -> 0.859 | Improved |
| add cross-region stability-score filtering of pseudo-labels (ablation) | GRAM · Dar es Salaam | mIoU: 0.818 -> 0.859 | Improved |
| use confidence-based filtering instead of stability filtering (ablation) | GRAM · Dar es Salaam | mIoU: 0.859 -> 0.463 | Hurt |
| use temporal consistency filtering instead of cross-expert stability (ablation) | GRAM · Dar es Salaam | mIoU: 0.859 -> 0.837 | Hurt |
| confidence filtering compared with no filtering (ablation) | GRAM · Dar es Salaam | mIoU: 0.818 -> 0.463 | Hurt |
| use entropy minimization TTA (TENT) instead of no adaptation | Vanilla Source · Dar es Salaam | mIoU: 0.681 -> 0.691 | No clear effect |
| add cross-region consistency sampling to BeCoTTA's MoE adapters | BeCoTTA · three African cities | mIoU: Dar es Salaam 0.741 vs GRAM 0.859; Maputo 0.904 vs GRAM 0.907 | Mixed |
Closest papers by meaning
- UV-SAM: Adapting Segment Anything Model for Urban Village Identification (AAAI 2024 · similarity 0.84)
- Combining Deep Learning and Street View Imagery to Map Smallholder Crop Types (AAAI 2024 · similarity 0.84)
- ROS-SAM: High-Quality Interactive Segmentation for Remote Sensing Moving Object (CVPR 2025 · similarity 0.84)
- UrbanVLP: Multi-Granularity Vision-Language Pretraining for Urban Socioeconomic Indicator Prediction (AAAI 2025 · similarity 0.84)
- Fields of The World: A Machine Learning Benchmark Dataset for Global Agricultural Field Boundary Segmentation (AAAI 2025 · similarity 0.84)
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.83)
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.
