Research map / Detection and segmentation / CNN features
GraphFPN: Graph Feature Pyramid Network for Object Detection
Gangming Zhao, Weifeng Ge, Yizhou Yu
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 |
|---|---|---|---|
| add first group of contextual layers (CGL-1) (ablation) | GraphFPN · MS-COCO 2017 val | AP: 38.2 -> 39.1 | Improved |
| add second group of contextual layers (CGL-2) (ablation) | GraphFPN · MS-COCO 2017 val | AP: 38.7 -> 39.1 | Improved |
| add hierarchical layers to contextual-only layers (ablation) | GraphFPN · MS-COCO 2017 val | AP: 37.2 -> 39.1 | Improved |
| add contextual layers to hierarchical-only layers (ablation) | GraphFPN · MS-COCO 2017 val | AP: 36.2 -> 39.1 | Improved |
| add spatial self-attention (ablation) | GraphFPN · MS-COCO 2017 val | AP: 37.8 -> 39.1 | Improved |
| add local channel-wise attention (ablation) | GraphFPN · MS-COCO 2017 val | AP: 37.9 -> 39.1 | Improved |
| add local channel self-attention (ablation) | GraphFPN · MS-COCO 2017 val | AP: 37.6 -> 39.1 | Improved |
| add both local channel attentions (ablation) | GraphFPN · MS-COCO 2017 val | AP: 37.1 -> 39.1 | Improved |
| increase layers per group from 1 to 3 (ablation) | GraphFPN · MS-COCO 2017 val | AP: 36.1 -> 39.1 | Improved |
| increase layers per group from 3 to 5 (ablation) | GraphFPN · MS-COCO 2017 val | AP: 39.1 -> 37.1 | Hurt |
| add GraphFPN to Faster R-CNN+FPN | Faster R-CNN+FPN · MS-COCO 2017 test-dev | AP: 36.2 -> 43.7 | Improved |
| add multi-scale training and augmented head to GraphFPN | GraphFPN · MS-COCO 2017 test-dev | AP: 42.1 -> 43.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- ZigZagNet — ZigZagNet: Fusing Top-Down and Bottom-Up Context for Object Segmentation (CVPR 2019) · better on 2 of 2 shared results
- Deformable DETR — Deformable DETR: Deformable Transformers for End-to-End Object Detection (ICLR 2021) · better on 1 of 1 shared result
- Faster R-CNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Feature Pyramid Networks for Object Detection (CVPR 2017 · similarity 0.93)
- M2Det: A Single-Shot Object Detector Based on Multi-Level Feature Pyramid Network (AAAI 2019 · similarity 0.91)
- Parallel Feature Pyramid Network for Object Detection (ECCV 2018 · similarity 0.90)
- AugFPN: Improving Multi-Scale Feature Learning for Object Detection (CVPR 2020 · similarity 0.89)
- You Should Look at All Objects (ECCV 2022 · similarity 0.89)
- Construct Effective Geometry Aware Feature Pyramid Network for Multi-Scale Object Detection (AAAI 2022 · similarity 0.89)
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.
