Research map / Detection and segmentation / Open-set and few-shot detection
Distillation Using Oracle Queries for Transformer-Based Human-Object Interaction 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 DOQ and CCS to QPIC | QPIC · HICO-DET DT, full | mAP: 29.07 -> 31.55 | Helped |
| Apply DOQ and CCS to QPIC | QPIC · HOI-A | mAP: 74.10 -> 76.87 | Helped |
| Apply DOQ and CCS to QPIC | QPIC · V-COCO Scenario 1 | mAProle: 58.8 -> 63.5 | Helped |
| Add a teacher network with shared parameters, without distillation loss or CCS (ablation) | QPIC · HICO-DET DT, full | mAP: 29.07 -> 29.63 | Helped |
| Add decoder-embedding cosine distillation (ablation) | QPIC with MTL · HICO-DET DT, full | mAP: 29.63 -> 30.28 | Helped |
| Add attention-map KL distillation (ablation) | QPIC with MTL and Lcos · HICO-DET DT, full | mAP: 30.28 -> 30.41 | Helped |
| Add Context-Consistent Stitching (ablation) | QPIC baseline · HICO-DET DT, full | mAP: 29.07 -> 30.76 | Helped |
| Apply DOQ and CCS | HOTR · HICO-DET DT, full | mAP: 23.46 -> 25.97 | Helped |
| Apply DOQ and CCS | CDN-S · HICO-DET DT, full | mAP: 31.44 -> 33.28 | Helped |
| Use relative box positions and box areas in addition to absolute positions for oracle queries (ablation) | DOQ · HICO-DET DT, full | mAP: 30.12 -> 30.41 | Helped |
| Initialize decoder embeddings with CLIP object-category embeddings instead of zero vectors (ablation) | DOQ · HICO-DET DT, full | mAP: 29.70 -> 30.41 | Helped |
| Use object-category embeddings rather than verb-category label vectors for initial decoder embeddings (ablation) | DOQ · HICO-DET DT, full | mAP: 30.10 -> 30.41 | Helped |
Closest papers by meaning
- QueryCraft: Transformer-Guided Query Initialization for Enhanced Human-Object Interaction Detection (AAAI 2026 · similarity 0.91)
- Human-Object Interaction Detection via Disentangled Transformer (CVPR 2022 · similarity 0.91)
- QPIC: Query-Based Pairwise Human-Object Interaction Detection With Image-Wide Contextual Information (CVPR 2021 · similarity 0.90)
- HOTR: End-to-End Human-Object Interaction Detection With Transformers (CVPR 2021 · similarity 0.90)
- Toward Open-Set Human Object Interaction Detection (AAAI 2024 · similarity 0.88)
- HOIMamba: Efficient Mamba-based Disentangled Progressive Learning for HOI 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.