Research map / Detection and segmentation / Open-set and few-shot detection
RegFormer: Transferable Relational Grounding for Efficient Weakly-Supervised Human-Object Interaction Detection
Jihwan Park, Chanhyeong Yang, Jinyoung Park, Taehoon Song, Hyunwoo J. Kim
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 sequential HO->I decoding to ML-Decoder (ablation) | ML-Decoder (DINO-S) · HICO | mAP: 52.6 -> 53.7 | Improved |
| add sequential HO->I decoding to ML-Decoder (ablation) | ML-Decoder (DINO-S) · HICO-DET Full | mAP: 17.49 -> 17.63 | No clear effect |
| add spatially grounded pairwise query (ablation) | RegFormer (DINO-S) · HICO-DET Full | mAP: 17.63 -> 22.08 | Improved |
| add spatially grounded pairwise query (ablation) | RegFormer (DINO-S) · HICO | mAP: 53.7 -> 54.4 | Improved |
| add interactiveness scoring to HO->I + SG (ablation) | RegFormer (DINO-S) · HICO-DET Full | mAP: 22.08 -> 30.01 | Improved |
| add interactiveness scoring to HO->I + SG (ablation) | RegFormer (DINO-S) · HICO | mAP: 54.4 -> 57.6 | Improved |
| use both local and masked global interactiveness vs only one (ablation) | RegFormer (DINO-S+CLIP-B) · HICO-DET Full | mAP: 23.44 -> 30.01 | Improved |
| use stronger vision encoder (CLIP-RN50 to DINO-B) | RegFormer (Faster R-CNN) · HICO-DET Full | mAP: 25.08 -> 33.33 | Improved |
| use stronger detector H-DETR instead of DETR | RegFormer (DINO-B) · HICO-DET Full | mAP: 32.90 -> 38.14 | Improved |
| replace Weakly HOI-CLIP with RegFormer at same CLIP-RN50 backbone | Faster R-CNN · HICO-DET Full | mAP: 22.89 -> 25.08 | Improved |
| RegFormer vs OpenCat in zero-shot RF-UC | RegFormer · HICO-DET RF-UC Unseen | mAP: 21.46 -> 31.53 | Improved |
| weakly supervised RegFormer vs fully supervised HOICLIP | RegFormer (H-DETR, DINO-B) · HICO-DET Full | mAP: 34.69 -> 38.14 | Improved |
Closest papers by meaning
- LINK: Learning Instance-level Knowledge from Vision-Language Models for Human-Object Interaction Detection (ICLR 2026 · similarity 0.91)
- Weakly-supervised HOI Detection via Prior-guided Bi-level Representation Learning (ICLR 2023 · similarity 0.91)
- Learning Transferable Human-Object Interaction Detector With Natural Language Supervision (CVPR 2022 · similarity 0.90)
- Exploring Structure-Aware Transformer Over Interaction Proposals for Human-Object Interaction Detection (CVPR 2022 · similarity 0.90)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 2024 · similarity 0.89)
- Human-Object Interaction Detection Collaborated with Large Relation-driven Diffusion Models (NeurIPS 2024 · similarity 0.88)
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.
