Research map / Detection and segmentation / CNN features
Dogfight: Detecting Drones From Drones Videos
Check 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 focal loss and Distance-IOU loss to cross-entropy (ablation) | Proposed stage-1 · NPS-Drones | F1 score: 0.85 -> 0.86 | Helped |
| Add focal loss and Distance-IOU loss to cross-entropy (ablation) | Proposed stage-1 · NPS-Drones | Recall: 0.82 -> 0.84 | Helped |
| Add channel-wise attention (ablation) | Losses (focal + Distance-IOU) · NPS-Drones | F1 score: 0.86 -> 0.87 | Helped |
| Add channel-wise attention (ablation) | Losses (focal + Distance-IOU) · NPS-Drones | Recall: 0.84 -> 0.86 | Helped |
| Add pixel-wise attention (ablation) | Losses (focal + Distance-IOU) · NPS-Drones | F1 score: 0.86 -> 0.87 | Helped |
| Add pixel-wise attention (ablation) | Losses (focal + Distance-IOU) · NPS-Drones | Precision: 0.88 -> 0.89 | Helped |
| Add stage-2 motion-based candidate discovery and spatio-temporal verification (ablation) | Stage-1 · NPS-Drones | F1 score: 0.90 -> 0.92 | Helped |
| Add stage-2 motion-based candidate discovery and spatio-temporal verification (ablation) | Stage-1 · NPS-Drones | Recall: 0.88 -> 0.91 | Helped |
| Use multiple frames in a single-stage approach (ablation) | Proposed approach · NPS-Drones and FL-Drones | F1 score: below F1-score=40 | Hurt |
| Replace bilinear interpolation with patch or feature-map superresolution (ablation) | Stage-2 cuboid feature resizing · NPS-Drones and FL-Drones | detection performance | No clear effect |
| Use the two-stage segmentation and spatio-temporal method rather than the strongest reported baseline | Drone detection · NPS-Drones | F1 score: 0.86 -> 0.92 | Helped |
| Use the two-stage segmentation and spatio-temporal method rather than the strongest reported baseline | Drone detection · FL-Drones | F1 score: 0.71 -> 0.80 | Helped |
Closest papers by meaning
- PointSR: Self-Regularized Point Supervision for Drone-View Object Detection (CVPR 2025 · similarity 0.85)
- UAV3D: A Large-scale 3D Perception Benchmark for Unmanned Aerial Vehicles (NeurIPS 2024 · similarity 0.83)
- RemDet: Rethinking Efficient Model Design for UAV Object Detection (AAAI 2025 · similarity 0.82)
- Drones Help Drones: A Collaborative Framework for Multi-Drone Object Trajectory Prediction and Beyond (NeurIPS 2024 · similarity 0.82)
- MMOT: The First Challenging Benchmark for Drone-based Multispectral Multi-Object Tracking (NeurIPS 2025 · similarity 0.82)
- Multiview Aerial Visual RECognition (MAVREC): Can Multi-view Improve Aerial Visual Perception? (CVPR 2024 · similarity 0.82)
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.