Research map / Detection and segmentation / Open-set and few-shot detection
Miss Detection vs. False Alarm: Adversarial Learning for Small Object Segmentation in Infrared Images
Huan Wang, Luping Zhou, Lei Wang
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 |
|---|---|---|---|
| use two adversarial generators for MD and FA instead of single CAN8 with combined objective (ablation) | MDvsFA-cGAN · Single | F-measure: 0.31 -> 0.60 | Improved |
| use two adversarial generators instead of single UCAN64 with combined objective (ablation) | MDvsFA-cGAN · Single | F-measure: 0.39 -> 0.60 | Improved |
| use two adversarial generators instead of single CAN8 with combined objective (ablation) | MDvsFA-cGAN · AllSeqs | F-measure: 0.20 -> 0.27 | Improved |
| use two adversarial generators instead of single UCAN64-cGAN (ablation) | MDvsFA-cGAN · AllSeqs | F-measure: 0.22 -> 0.27 | Improved |
| use two adversarial generators instead of single UCAN64-cGAN (ablation) | MDvsFA-cGAN · Single | F-measure: 0.40 -> 0.60 | Improved |
| use different architectures (CAN8 and UCAN64) for the two generators instead of both CAN8 (ablation) | MDvsFA-cGAN · AllSeqs | F-measure: 0.17 -> 0.27 | Improved |
| use different architectures for the two generators instead of both CAN8 (ablation) | MDvsFA-cGAN · Single | F-measure: 0.37 -> 0.60 | Improved |
| use different architectures for the two generators instead of both UCAN64 (ablation) | MDvsFA-cGAN · AllSeqs | F-measure: 0.20 -> 0.27 | Improved |
| use different architectures for the two generators instead of both UCAN64 (ablation) | MDvsFA-cGAN · Single | F-measure: 0.50 -> 0.60 | Improved |
| vary lambda1 from 100 to 10 with lambda2=1 (ablation) | MDvsFA-cGAN · Single | F-measure: 0.60 -> 0.58 | No clear effect |
| vary lambda2 from 10 down to 0.1 with lambda1=100 (ablation) | MDvsFA-cGAN · Single | F-measure: 0.60 -> 0.60 | No clear effect |
| add adversarial MD/FA learning versus generic small-object segmentation FCN-MFB | MDvsFA-cGAN · Single | F-measure: 0.47 -> 0.60 | Improved |
Closest papers by meaning
- IRSAM: Advancing Segment Anything Model for Infrared Small Target Detection (ECCV 2024 · similarity 0.86)
- Seeing Through the Noise: Improving Infrared Small Target Detection and Segmentation from Noise Suppression Perspective (CVPR 2026 · similarity 0.85)
- Rethinking Evaluation of Infrared Small Target Detection (NeurIPS 2025 · similarity 0.85)
- UNIP: Rethinking Pre-trained Attention Patterns for Infrared Semantic Segmentation (ICLR 2025 · similarity 0.84)
- ASAM: Boosting Segment Anything Model with Adversarial Tuning (CVPR 2024 · similarity 0.84)
- Semi-supervised Infrared Small Target Detection with Thermodynamic-Inspired Uneven Perturbation and Confidence Adaptation (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.
