Research map / Detection and segmentation / Weakly supervised detection and segmentation
Mapping Degeneration Meets Label Evolution: Learning Infrared Small Target Detection With Single Point Supervision
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 LESPS to DNA-Net under centroid-point supervision | DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K) | IoU (×10²): 5.12 -> 57.34 | Helped |
| Add LESPS to DNA-Net under coarse-point supervision | DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K) | IoU (×10²): 2.96 -> 56.18 | Helped |
| Add LESPS to DNA-Net under centroid-point supervision | DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K) | Pd (×10²): 89.19 -> 91.87 | Helped |
| Use coarse rather than centroid point supervision with LESPS | DNA-Net + LESPS · Average (SIRST, NUDT-SIRST, IRSTD-1K) | Pd (×10²): 91.87 vs. 91.49 | No clear effect |
| Replace fixed pseudo labels with dynamic LESPS label updates | DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K), centroid supervision | IoU (×10²): 28.89 -> 57.34 | Helped |
| Compute negative loss on rand₅ rather than all background points (ablation) | DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K), centroid supervision | IoU (×10²): 5.12 -> 24.80 | Helped |
| Compute negative loss on rand₅ rather than all background points (ablation) | DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K), centroid supervision | Fa (×10³): 0.01 -> 0.34 | Hurt |
| Start label evolution at T_loss=50 rather than the selected T_loss=10 (ablation) | LESPS · Parameter analysis | IoU: 0.04 vs. 0.54 | Hurt |
| Start label evolution at T_loss=2 rather than the selected T_loss=10 (ablation) | LESPS · Parameter analysis | IoU: 0.21 vs. 0.54 | Hurt |
| Increase label-update frequency from every 5 epochs to every 2 epochs (ablation) | LESPS · Parameter analysis | IoU | Helped |
| Use rand₅ background points instead of the paper's all-background loss (ablation) | ACM, ALCNet, and DNA-Net · Average (SIRST, NUDT-SIRST, IRSTD-1K), centroid supervision | IoU (×10²): 3.95 -> 13.39; 4.08 -> 11.87; 5.12 -> 24.80 | Helped |
Closest papers by meaning
- Semi-supervised Infrared Small Target Detection with Thermodynamic-Inspired Uneven Perturbation and Confidence Adaptation (AAAI 2025 · similarity 0.90)
- Rethinking Evaluation of Infrared Small Target Detection (NeurIPS 2025 · similarity 0.87)
- Pinwheel-shaped Convolution and Scale-based Dynamic Loss for Infrared Small Target Detection (AAAI 2025 · similarity 0.87)
- Rethinking Infrared Small Target Detection: A Foundation Driven Efficient Paradigm (NeurIPS 2026 · similarity 0.87)
- IRPruneDet: Efficient Infrared Small Target Detection via Wavelet Structure-Regularized Soft Channel Pruning (AAAI 2024 · similarity 0.86)
- Motion Prior Knowledge Learning with Homogeneous Language Descriptions for Moving Infrared Small Target Detection (AAAI 2025 · similarity 0.85)
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.