Research map / Detection and segmentation / Open-set and few-shot detection
STARLINC: Satellite Trail Artifact Removal using Inter-Frame Correlation
Shingeon Kim, Hyeyoon Lee, Dain Kwon, Kanghyun Choi, SunJong Park, Mi-Ryang Kim, Jeong-Eun Lee, Jinho Lee
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 inter-frame differential map (ablation) | segmentation U-Net · 7DT Orion segmentation | mIoU: 0.000 -> 0.410 | Improved |
| add Grad-CAM heatmap on top of differential map (ablation) | STARLINC segmentation · 7DT Orion segmentation | mIoU: 0.410 -> 0.485 | Improved |
| add Grad-CAM heatmap without differential map (ablation) | segmentation U-Net · 7DT Orion segmentation | mIoU: 0.000 -> 0.198 | Improved |
| use SSIM instead of RMSE for differential map (ablation) | STARLINC · 7DT Orion segmentation | mIoU: 0.469 -> 0.485 | Improved |
| use SSIM instead of ZOGY for differential map (ablation) | STARLINC · 7DT Orion segmentation | mIoU: 0.392 -> 0.485 | Improved |
| replace geometric trail renderer with PSF-aware renderer (ablation) | STARLINC · 7DT Orion segmentation | mIoU: 0.485 -> 0.539; precision 0.815 -> 0.748 | Mixed |
| use 16-bit input instead of 8-bit (ablation) | STARLINC · 7DT Orion segmentation | mIoU: 0.485 -> 0.450 | Hurt |
| raise synthetic fraction among trail images from 57.3% to 95.1% (ablation) | STARLINC · 7DT Orion segmentation | mIoU: 0.485 -> 0.504 | No clear effect |
| scale U-Net from 3.67M to 20.61M parameters (ablation) | STARLINC · 7DT Orion segmentation | mIoU: 0.475 -> 0.473 | No clear effect |
| add synthetic trail generation (no differential map) (ablation) | trail classifier · 7DT Orion trail classification | F1: 0.314 -> 0.381 | Improved |
| add inter-frame differential map (no synthetic trails) (ablation) | trail classifier · 7DT Orion trail classification | F1: 0.314 -> 0.779 | Improved |
| combine synthetic trails and differential map (ablation) | trail classifier · 7DT Orion trail classification | F1: 0.779 -> 0.959 | Improved |
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.
- DeepLSD — DeepLSD: Line Segment Detection and Refinement With Deep Image Gradients (CVPR 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
- FluxFlow: Conservative Flow-Matching for Astronomical Image Super-Resolution (NeurIPS 2026 · similarity 0.81)
- AllClear: A Comprehensive Dataset and Benchmark for Cloud Removal in Satellite Imagery (NeurIPS 2024 · similarity 0.80)
- Satellite to GroundScape - Large-scale Consistent Ground View Generation from Satellite Views (CVPR 2025 · similarity 0.80)
- COSMIC: Compress Satellite Image Efficiently via Diffusion Compensation (NeurIPS 2024 · similarity 0.80)
- Deep Linear Array Pushbroom Image Restoration: A Degradation Pipeline and Jitter-Aware Restoration Network (AAAI 2024 · similarity 0.80)
- STAR: A Benchmark for Astronomical Star Fields Super-Resolution (NeurIPS 2025 · similarity 0.79)
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.
