Research map / Detection and segmentation / Open-set and few-shot detection
GEN-VLKT: Simplify Association and Enhance Interaction Understanding for HOI 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 |
|---|---|---|---|
| Use GEN-VLKT_l rather than QPIC-R101 | HICO-Det test set · HICO-Det Default Full | mAP: 29.90 -> 34.95 | Helped |
| Use GEN-VLKT_l rather than QPIC-R50 | V-COCO · Scenario 1 | role AP: 58.8 -> 63.58 | Helped |
| Use GEN-VLKT_s rather than the no-VLKT baseline | HICO-Det · RF-UC unseen categories | mAP: 12.52 -> 21.36 | Helped |
| Replace the verb classifier with an HOI triplet classifier (ablation) | GEN · HICO-Det default Full | mAP: 31.88 -> 30.96 | Hurt |
| Add p-GE with independent human and object queries (ablation) | GEN Base-verb · HICO-Det default Full | mAP: 31.23 -> 31.88 | Helped |
| Add i-GE to guide interaction queries (ablation) | GEN Base-verb · HICO-Det default Full | mAP: 30.83 -> 31.88 | Helped |
| Initialize the interaction classifier with CLIP text embeddings (ablation) | Base-triplet · HICO-Det default Rare | mAP: 22.28 -> 26.08 | Helped |
| Add CLIP text initialization for the object classifier (ablation) | Base-triplet with interaction text · HICO-Det default Full | mAP: 31.71 -> 32.09 | Helped |
| Add CLIP visual embedding mimic loss (ablation) | Base-triplet with text-driven classifiers · HICO-Det default Full | mAP: 32.09 -> 33.75 | Helped |
| Use L2 rather than L1 alone for visual mimic (ablation) | GEN-VLKT · HICO-Det default Full | mAP: 33.75 -> 32.41 | Hurt |
| Combine L1 and L2 mimic losses rather than using L1 alone (ablation) | GEN-VLKT · HICO-Det default Full | mAP: 33.75 -> 33.10 | Hurt |
Closest papers by meaning
- Exploring the Potential of Large Foundation Models for Open-Vocabulary HOI Detection (CVPR 2024 · similarity 0.90)
- Toward Open-Set Human Object Interaction Detection (AAAI 2024 · similarity 0.89)
- Open-World Human-Object Interaction Detection via Multi-modal Prompts (CVPR 2024 · similarity 0.89)
- Discovering Syntactic Interaction Clues for Human-Object Interaction Detection (CVPR 2024 · similarity 0.88)
- Dual-Prior Augmented Decoding Network for Long Tail Distribution in HOI Detection (AAAI 2024 · similarity 0.88)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 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.