Research map / Detection and segmentation / CNN features
Scale-Aware Automatic Augmentation for Object Detection
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 |
|---|---|---|---|
| apply the searched scale-aware augmentation policy | RetinaNet ResNet-50 · MS COCO val2017 | AP: 38.2 -> 41.3 | Helped |
| apply the searched scale-aware augmentation policy | Faster R-CNN ResNet-50 · MS COCO val2017 | AP: 39.1 -> 41.8 | Helped |
| apply the searched scale-aware augmentation policy | FCOS ResNet-101 · MS COCO val2017 | AP: 41.8 -> 44.0 | Helped |
| apply the searched policy | Faster R-CNN ResNet-50 · PASCAL VOC | mAP: 78.6 -> 81.6 | Helped |
| add image-level zoom augmentations (ablation) | RetinaNet ResNet-50 · MS COCO val2017 | AP: 38.2 -> 40.1 | Helped |
| add non-scale-aware box-level augmentations (ablation) | RetinaNet ResNet-50 with image-level augmentation · MS COCO val2017 | AP: 40.1 -> 40.6 | Helped |
| make box-level augmentation areas scale-aware (ablation) | RetinaNet ResNet-50 with image- and box-level augmentation · MS COCO val2017 | AP: 40.6 -> 41.3 | Helped |
| use Pareto Scale Balance rather than proxy accuracy for search (ablation) | RetinaNet ResNet-50 policy search · MS COCO val2017 | AP: 40.0 -> 41.3 | Helped |
| add the accuracy-drop penalty to scale-loss standard deviation (ablation) | RetinaNet ResNet-50 policy search · MS COCO val2017 | AP: 40.7 -> 41.3 | Helped |
| remove context pixels | Faster R-CNN ResNet-101 · MS COCO val2017, small objects | AP_s: 25.2 -> 18.0 | Hurt |
| remove context pixels | Faster R-CNN ResNet-101 · MS COCO val2017, large objects | AP_l: 53.0 -> 56.1 | Helped |
| remove context pixels | RetinaNet · MS COCO val2017 | AP_s: 23.3 -> 16.7 | Hurt |
Closest papers by meaning
- RandAugment: Practical Automated Data Augmentation with a Reduced Search Space (NeurIPS 2020 · similarity 0.89)
- Robust and Accurate Object Detection via Adversarial Learning (CVPR 2021 · similarity 0.86)
- ScaleKD: Distilling Scale-Aware Knowledge in Small Object Detector (CVPR 2023 · similarity 0.86)
- Your Image is My Video: Reshaping the Receptive Field via Image-To-Video Differentiable AutoAugmentation and Fusion (CVPR 2024 · similarity 0.85)
- Size-invariance Matters: Rethinking Metrics and Losses for Imbalanced Multi-object Salient Object Detection (ICML 2024 · similarity 0.84)
- PhysAug: A Physical-guided and Frequency-based Data Augmentation for Single-Domain Generalized Object Detection (AAAI 2025 · similarity 0.84)
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.