Research map / Detection and segmentation / Open-set and few-shot detection
On Model Calibration for Long-Tailed Object Detection and Instance Segmentation
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 |
|---|---|---|---|
| Apply NorCal to RFS | RFS, R-50 · LVIS v1 instance segmentation | AP: 22.58 -> 25.22 | Helped |
| Apply NorCal to RFS | RFS, R-50 · LVIS v1 instance segmentation | AP_f: 28.55 -> 28.98 | Helped |
| Apply NorCal to EQL | EQL, R-50 · LVIS v1 instance segmentation | AP_f: 27.20 -> 27.10 | Hurt |
| Apply NorCal to Faster R-CNN | Faster R-CNN, R-50 · MSCOCO object detection | AP: 37.93 -> 37.96 | No clear effect |
| Apply class score normalization (ablation) | NorCal with CDT and probability scaling · LVIS v1 instance segmentation | AP: 24.50 -> 25.22 | Helped |
| Apply class score normalization (ablation) | NorCal with CDT and exponential scaling · LVIS v1 instance segmentation | AP: 17.52 -> 25.22 | Helped |
| Scale probabilities instead of logits (ablation) | NorCal with CDT and normalization · LVIS v1 instance segmentation | AP: 24.77 -> 25.22 | Helped |
| Use ENS instead of CDT (ablation) | NorCal with probability scaling and normalization · LVIS v1 instance segmentation | AP: 25.22 -> 24.85 | Hurt |
| Apply NorCal to RFS | RFS, R-50 · LVIS v1 instance segmentation | AR: 30.61 -> 36.10 | Helped |
| Keep the background logit uncalibrated rather than reducing its exponential (ablation) | NorCal · LVIS v1 instance segmentation | AP | Helped |
Closest papers by meaning
- Fractal Calibration for Long-tailed Object Detection (CVPR 2025 · similarity 0.89)
- Image-Level or Object-Level? A Tale of Two Resampling Strategies for Long-Tailed Detection (ICML 2021 · similarity 0.89)
- SimLTD: Simple Supervised and Semi-Supervised Long-Tailed Object Detection (CVPR 2025 · similarity 0.87)
- Long-tailed Object Detection Pretraining: Dynamic Rebalancing Contrastive Learning with Dual Reconstruction (NeurIPS 2024 · similarity 0.87)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.85)
- Object Detection Benchmarks are Incomplete: The Role of Label Errors and Annotation Uncertainty (NeurIPS 2026 · similarity 0.85)
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.