Research map / Detection and segmentation / CNN features
AugFPN: Improving Multi-Scale Feature Learning 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 |
|---|---|---|---|
| Replace FPN with AugFPN | Faster R-CNN, ResNet-50 · COCO test-dev | AP: 36.5 -> 38.8 | Helped |
| Replace FPN with AugFPN | Faster R-CNN, MobileNet-v2 · COCO test-dev | AP: 32.6 -> 34.2 | Helped |
| Replace FPN with AugFPN | RetinaNet, ResNet-50 · COCO test-dev | AP: 35.9 -> 37.5 | Helped |
| Replace FPN with AugFPN | FCOS, ResNet-50 · COCO test-dev | AP: 37.0 -> 37.9 | Helped |
| Add Consistent Supervision (ablation) | Faster R-CNN, ResNet-50-FPN · COCO val2017 | AP: 36.3 -> 37.2 | Helped |
| Add Residual Feature Augmentation (ablation) | Faster R-CNN, ResNet-50-FPN · COCO val2017 | AP: 36.3 -> 37.3 | Helped |
| Add Soft RoI Selection (ablation) | Faster R-CNN, ResNet-50-FPN · COCO val2017 | AP: 36.3 -> 37.1 | Helped |
| Apply supervision at all feature levels instead of only the assigned level (ablation) | Consistent Supervision, Faster R-CNN ResNet-50-FPN · COCO val2017 | AP: 36.7 -> 37.2 | Helped |
| Use global max pooling for context instead of the baseline (ablation) | Residual Feature Augmentation · COCO val2017 | AP: 36.3 -> 34.5 | Hurt |
| Use ratio-invariant adaptive average pooling instead of global average pooling (ablation) | Residual Feature Augmentation · COCO val2017 | AP: 36.8 -> 37.1 | Helped |
| Add Adaptive Spatial Fusion to ratio-invariant adaptive pooling (ablation) | Residual Feature Augmentation · COCO val2017 | AP: 37.1 -> 37.3 | Helped |
| Replace ratio-invariant adaptive pooling with PSP pooling (ablation) | Residual Feature Augmentation · COCO val2017 | AP: 37.3 -> 37.0 | Hurt |
Closest papers by meaning
- M2Det: A Single-Shot Object Detector Based on Multi-Level Feature Pyramid Network (AAAI 2019 · similarity 0.91)
- Feature Pyramid Networks for Object Detection (CVPR 2017 · similarity 0.91)
- Construct Effective Geometry Aware Feature Pyramid Network for Multi-Scale Object Detection (AAAI 2022 · similarity 0.90)
- FAS-Net: Construct Effective Features Adaptively for Multi-Scale Object Detection (AAAI 2020 · similarity 0.89)
- NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detection (CVPR 2019 · similarity 0.89)
- HS-FPN: High Frequency and Spatial Perception FPN for Tiny Object Detection (AAAI 2025 · similarity 0.87)
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.