Research map / Video / Video segmentation and tracking
Video-based Human-Object Interaction Detection from Tubelet Tokens
Read the paperCheck 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 |
|---|---|---|---|
| replace patch-token processing with TUTOR | VidHOI · VidHOI Full | mAP: 22.92 -> 26.92 | Helped |
| add token abstraction (ablation) | TUTOR · VidHOI static HOI (S) | mAP: 9.67 -> 16.42 | Helped |
| add token linking (ablation) | TUTOR · VidHOI temporal HOI (T) | mAP: 2.30 -> 8.28 | Helped |
| use irregular rather than regular windows for token abstraction (ablation) | TUTOR · VidHOI static HOI (S) | mAP: 28.17 -> 32.21 | Helped |
| use irregular rather than regular windows for token abstraction (ablation) | TUTOR · VidHOI temporal HOI (T) | mAP: 18.24 -> 21.28 | Helped |
| fuse four neighboring patch tokens after each Transformer layer (ablation) | ViT-like framework · VidHOI static HOI (S) | mAP: 27.64 -> 28.45 | Helped |
| replace nms-one-hot with direct Gumbel-Softmax assignment weights (ablation) | TUTOR token linking · VidHOI static HOI (S) | mAP: 32.21 -> 28.81 | Hurt |
| replace nms-one-hot with direct Gumbel-Softmax assignment weights (ablation) | TUTOR token linking · VidHOI temporal HOI (T) | mAP: 21.28 -> 18.11 | Hurt |
| remove one-to-one suppression from the token assignment (ablation) | TUTOR token linking · VidHOI static HOI (S) | mAP: 32.21 -> 30.64 | Hurt |
| replace clustering agglomeration with k-means (ablation) | TUTOR token agglomeration · VidHOI static HOI (S) | mAP: 32.21 -> 26.71 | Hurt |
| replace irregular windows with regular windows for agglomeration (ablation) | TUTOR token agglomeration · VidHOI static HOI (S) | mAP: 32.21 -> 28.82 | Hurt |
| increase agglomerated token dimension from C to 2C (ablation) | TUTOR token agglomeration · VidHOI static HOI (S) | mAP: 29.34 -> 32.21 | Helped |
Closest papers by meaning
- LINK: Learning Instance-level Knowledge from Vision-Language Models for Human-Object Interaction Detection (ICLR 2026 · similarity 0.89)
- Learning Transferable Human-Object Interaction Detector With Natural Language Supervision (CVPR 2022 · similarity 0.88)
- Distillation Using Oracle Queries for Transformer-Based Human-Object Interaction Detection (CVPR 2022 · similarity 0.88)
- QPIC: Query-Based Pairwise Human-Object Interaction Detection With Image-Wide Contextual Information (CVPR 2021 · similarity 0.88)
- TubeR: Tubelet Transformer for Video Action Detection (CVPR 2022 · similarity 0.88)
- What To Look at and Where: Semantic and Spatial Refined Transformer for Detecting Human-Object Interactions (CVPR 2022 · 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.