Research map / Detection and segmentation / CNN features
VarifocalNet: An IoU-Aware Dense Object Detector
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 |
|---|---|---|---|
| Learn an IoU-aware classification score to rank detections | FCOS+ATSS · COCO val2017 | AP: 39.2 -> 74.7 | Helped |
| Add Varifocal Loss instead of focal loss (ablation) | VFNet without Star Dconv or refinement · COCO val2017 | AP: 39.0 -> 40.1 | Helped |
| Add Star Dconv bounding-box representation (ablation) | VFNet with Varifocal Loss · COCO val2017 | AP: 40.1 -> 40.7 | Helped |
| Add bounding-box refinement (ablation) | VFNet with Varifocal Loss and Star Dconv · COCO val2017 | AP: 40.7 -> 41.6 | Helped |
| Weight positive-example loss by target q (ablation) | VFNet · COCO val2017 | AP: 41.2 -> 41.6 | Helped |
| Vary Varifocal Loss alpha and gamma across the reported settings (ablation) | VFNet · COCO val2017 | AP: 41.2 to 41.6 | No clear effect |
| Use Varifocal Loss instead of focal loss (ablation) | RetinaNet · COCO val2017 | AP: 36.5 -> 37.4 | Helped |
| Use Varifocal Loss instead of focal loss (ablation) | FoveaBox · COCO val2017 | AP: 36.3 -> 37.2 | Helped |
| Use Varifocal Loss instead of focal loss (ablation) | RepPoints · COCO val2017 | AP: 38.3 -> 39.7 | Helped |
| Use Varifocal Loss instead of focal loss (ablation) | ATSS · COCO val2017 | AP: 39.3 -> 40.2 | Helped |
| Use Varifocal Loss instead of Generalized Focal Loss (ablation) | VFNet · COCO val2017 | AP: 41.1 -> 41.6 | Helped |
| Use Varifocal Loss instead of focal loss (ablation) | VFNet · COCO val2017 | AP: 40.0 -> 41.6 | Helped |
Closest papers by meaning
- RankDetNet: Delving Into Ranking Constraints for Object Detection (CVPR 2021 · similarity 0.87)
- Precise Detection in Densely Packed Scenes (CVPR 2019 · similarity 0.86)
- M2Det: A Single-Shot Object Detector Based on Multi-Level Feature Pyramid Network (AAAI 2019 · similarity 0.86)
- Improving Object Localization With Fitness NMS and Bounded IoU Loss (CVPR 2018 · similarity 0.86)
- Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression (AAAI 2020 · similarity 0.86)
- Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection (CVPR 2021 · similarity 0.86)
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.