Research map / Detection and segmentation / Open-set and few-shot detection
Long-Tail Detection with Effective Class-Margins
Jang Hyun Cho, Philipp Krähenbühl
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 |
|---|---|---|---|
| replace cross-entropy with ECM Loss | Mask R-CNN R-50 1x · LVIS v1 val | mAP_segm: 21.7 -> 26.3 | Improved |
| replace Seesaw Loss with ECM Loss | Mask R-CNN R-50 1x · LVIS v1 val | AP_r: 16.4 -> 19.5 | Improved |
| replace Seesaw Loss with ECM Loss | Mask R-CNN R-50 2x · LVIS v1 val | mAP_segm: 26.7 -> 27.4 | Improved |
| replace Seesaw Loss with ECM Loss | Mask R-CNN R-101 2x · LVIS v1 val | mAP_segm: 28.1 -> 28.7 | Improved |
| replace Seesaw Loss with ECM Loss | Cascade Mask R-CNN R-101 · LVIS v1 val | mAP_segm: 30.1 -> 30.6 | Improved |
| replace Seesaw Loss with ECM Loss | Cascade Mask R-CNN Swin-B · LVIS v1 val | mAP_segm: 38.7 -> 39.7 | Improved |
| replace Focal Loss with ECM Loss | FCOS R-50 1x · LVIS v1 val | mAP_bbox: 22.0 -> 23.2 | Improved |
| replace Focal Loss with ECM Loss | ATSS R-50 1x · LVIS v1 val | AP_r: 8.6 -> 15.8 | Improved |
| replace Focal Loss with ECM Loss | ATSS R-101 1x · LVIS v1 val | mAP_bbox: 25.2 -> 26.5 | Improved |
| replace Varifocal Loss with ECM Loss | VarifocalNet R-50 1x · LVIS v1 val | mAP_bbox: 24.8 -> 25.7 | Improved |
| replace Focal Loss with ECM Loss under longer 2x schedule | FCOS R-50 · LVIS v1 val | mAP_bbox: 23.5 -> 24.4 | Improved |
| replace Focal Loss with ECM Loss under longer 2x schedule | ATSS R-50 · LVIS v1 val | mAP_bbox: 25.6 -> 26.1 | Improved |
Closest papers by meaning
- C2AM Loss: Chasing a Better Decision Boundary for Long-Tail Object Detection (CVPR 2022 · similarity 0.90)
- Boosting Long-tailed Object Detection via Step-wise Learning on Smooth-tail Data (ICCV 2023 · similarity 0.89)
- Adaptive Hierarchical Representation Learning for Long-Tailed Object Detection (CVPR 2022 · similarity 0.88)
- Pursuing Better Decision Boundaries for Long-Tailed Object Detection via Category Information Amount (ICLR 2025 · similarity 0.88)
- SimLTD: Simple Supervised and Semi-Supervised Long-Tailed Object Detection (CVPR 2025 · similarity 0.86)
- Difficulty-aware Balancing Margin Loss for Long-tailed Recognition (AAAI 2025 · 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.
