Research map / Detection and segmentation / CNN features
SCALoss: Side and Corner Aligned Loss for Bounding Box Regression
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 |
|---|---|---|---|
| Replace the regression loss with SCALoss | YOLOv3-tiny · COCO 2017 val | mAP: 18.9 -> >19.9 | Helped |
| Replace the regression loss with SCALoss | YOLOv3 · COCO 2017 val | mAP: 44.8 -> >45.3 | Helped |
| Replace the regression loss with SCALoss | SSD · PASCAL VOC 2007 test | mAP: 52.28 -> >53.45 | Helped |
| Replace the second-stage regression loss with SCALoss | Faster R-CNN · PASCAL VOC 2007 test | mAP: 50.85 -> >51.84 | Helped |
| Replace the regression loss with SCALoss | Faster R-CNN · LVIS 1.0 val | mAP: 16.7 -> >17.7 | Helped |
| Add Corner Distance to IoU (ablation) | SSD · PASCAL VOC 2007 test | mAP: 52.28 -> 52.97 | Helped |
| Add Center Distance to IoU (ablation) | SSD · PASCAL VOC 2007 test | mAP: 52.28 -> 52.65 | Helped |
| Use Corner Distance alone instead of IoU (ablation) | SSD · PASCAL VOC 2007 test | mAP: 52.28 -> 52.64 | Helped |
| Use Side Overlap alone instead of IoU (ablation) | SSD · PASCAL VOC 2007 test | mAP: +0.8 mAP | Helped |
| Combine Side Overlap with Corner Distance (ablation) | SSD · PASCAL VOC 2007 test | mAP: 52.28 -> 53.45 | Helped |
| Use Center Distance alone (ablation) | SSD · PASCAL VOC 2007 test | mAP: 19.17 | Hurt |
| Set SCALoss alpha to 0.2 instead of 0.5 (ablation) | SSD · PASCAL VOC 2007 test | mAP: 53.45 -> 53.23 | Hurt |
Closest papers by meaning
- Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression (AAAI 2020 · similarity 0.89)
- Bounding Box Stability against Feature Dropout Reflects Detector Generalization across Environments (ICLR 2024 · similarity 0.85)
- Union-over-Intersections: Object Detection beyond Winner-Takes-All (ICLR 2025 · similarity 0.84)
- Theoretically Achieving Continuous Representation of Oriented Bounding Boxes (CVPR 2024 · similarity 0.83)
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.83)
- ABBSPO: Adaptive Bounding Box Scaling and Symmetric Prior based Orientation Prediction for Detecting Aerial Image Objects (CVPR 2025 · 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.