Research map / Detection and segmentation / Open-set and few-shot detection
FGAHOI: Fine-Grained Anchors for Human-Object Interaction Detection
Shuailei Ma, Yuefeng Wang, Shanze Wang, Ying Jin Wei
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 |
|---|---|---|---|
| use stage-wise training instead of end-to-end (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 23.98 -> 29.94 | Improved |
| use stage-wise training instead of end-to-end (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Known Object Full | role mAP: 26.44 -> 32.48 | Improved |
| add multi-scale sampling to QAHOI (ablation) | QAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 27.67 -> 28.45 | Improved |
| add hierarchical spatial-aware merging (HSAM) (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 28.45 -> 29.60 | Improved |
| add task-aware merging (TAM) (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 28.45 -> 29.32 | Improved |
| combine HSAM and TAM (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 29.60 -> 29.94 | Improved |
| add HSAM (ablation) | FGAHOI baseline (Swin-Tiny) · HOI-SDC | role mAP: 21.18 -> 21.91 | Improved |
| add TAM (ablation) | FGAHOI baseline (Swin-Tiny) · HOI-SDC | role mAP: 21.18 -> 21.84 | Improved |
| add multi-scale sampling baseline over QAHOI (ablation) | QAHOI (Swin-Tiny) · HOI-SDC | role mAP: 19.55 -> 21.18 | Improved |
| increase sampling sizes from [1,3,5] to [5,7,9] (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 29.94 -> 29.65 | No clear effect |
| increase sampling sizes from [1,3,5] to [3,5,7] (ablation) | FGAHOI (Swin-Tiny) · HICO-DET Default Full | role mAP: 29.94 -> 29.72 | No clear effect |
| replace QAHOI with FGAHOI | QAHOI (Swin-Large) · HICO-DET Default Full | role mAP: 35.43 -> 37.18 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- HOTR — HOTR: End-to-End Human-Object Interaction Detection With Transformers (CVPR 2021) · better on 0 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
- Diagnosing Human-Object Interaction Detectors (IJCV 2025 · similarity 0.91)
- Exploiting Scene Graphs for Human-Object Interaction Detection (ICCV 2021 · similarity 0.91)
- Learning Human-Object Interaction as Groups (NeurIPS 2025 · similarity 0.90)
- Exploring the Potential of Large Foundation Models for Open-Vocabulary HOI Detection (CVPR 2024 · similarity 0.89)
- ContextHOI: Spatial Context Learning for Human-Object Interaction Detection (AAAI 2025 · similarity 0.89)
- An Image-like Diffusion Method for Human-Object Interaction Detection (CVPR 2025 · 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.
