Research map / Detection and segmentation / Open-set and few-shot detection
Image-Level or Object-Level? A Tale of Two Resampling Strategies for Long-Tailed Detection
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 |
|---|---|---|---|
| Combine image-level RFS with object-level memory replay instead of using RFS alone | Mask R-CNN, ResNet-50-FPN · LVIS v0.5 detection | APb: 24.3 -> 25.7 | Helped |
| Combine image-level RFS with object-level memory replay instead of using RFS alone | Mask R-CNN, ResNet-50-FPN · LVIS v0.5 detection | APr: 13.1 -> 17.2 | Helped |
| Use object-centric memory replay rather than image-level RFS | Mask R-CNN, ResNet-50-FPN · LVIS v0.5 detection | APb: 24.3 -> 24.4 | Helped |
| Use object-centric memory replay rather than image-level RFS | Mask R-CNN, ResNet-50-FPN · LVIS v0.5 detection | APr: 13.1 -> 10.4 | Hurt |
| Add EQL to RIO | RIO, ResNet-50 · LVIS v0.5 detection | APb: 25.7 -> 26.1 | Helped |
| Add EQL to RIO | RIO, ResNet-50 · LVIS v0.5 segmentation | APm: 26.0 -> 26.3 | Helped |
| Repeat frequent objects along with rare ones when repeating images | Image-level RFS · LVIS v0.5 training distribution | frequent object instance increase: 18% | Hurt |
| Combine image-level resampling with object-level resampling | RIO · LVIS v1.0 detection, ResNet-50 | APb: 23.3 -> 24.1 | Helped |
| Target classes with fewer than 30 samples rather than a broader or narrower threshold (ablation) | RIO memory bank · LVIS v0.5 detection | AP | Mixed |
| Retrieve 20 object features per eligible batch rather than varying the replay count (ablation) | RIO memory bank · LVIS v0.5 detection | AP | Mixed |
| Use memory replay instead of simply repeating current rare-object RoI features (ablation) | Object-centric resampling · LVIS v0.5 detection | AP | Helped |
Closest papers by meaning
- Adaptive Hierarchical Representation Learning for Long-Tailed Object Detection (CVPR 2022 · similarity 0.89)
- On Model Calibration for Long-Tailed Object Detection and Instance Segmentation (NeurIPS 2021 · similarity 0.89)
- Long-tailed Object Detection Pretraining: Dynamic Rebalancing Contrastive Learning with Dual Reconstruction (NeurIPS 2024 · similarity 0.88)
- SimLTD: Simple Supervised and Semi-Supervised Long-Tailed Object Detection (CVPR 2025 · similarity 0.88)
- OFBD: Object-Focused Background Debiasing for Long-Tailed Learning (NeurIPS 2026 · similarity 0.86)
- Decoupled Optimisation for Long-Tailed Visual Recognition (AAAI 2024 · 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.