Research map / Vision-language and multimodal / Robot action and navigation
YouRefIt: Embodied Reference Understanding With Language and Gesture
Yixin Chen, Qing Li, Deqian Kong, Yik Lun Kei, Song-Chun Zhu, Tao Gao, Yixin Zhu, Siyuan Huang
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 gestural cues (original images instead of inpainted) (ablation) | FAOA · YouRefIt Image ERU test (IoU=0.25, all sizes) | accuracy (%): 23.4 -> 44.5 | Improved |
| add gestural cues (original images instead of inpainted) (ablation) | ReSC · YouRefIt Image ERU test (IoU=0.25, all sizes) | accuracy (%): 34.3 -> 49.2 | Improved |
| add gestural cues (original images instead of inpainted) (ablation) | ReSC · YouRefIt Image ERU test (IoU=0.5, all sizes) | accuracy (%): 25.7 -> 34.9 | Improved |
| add language input (ablation) | Ours (gesture features) · YouRefIt Image ERU test (IoU=0.25, all sizes) | accuracy (%): 41.4 -> 54.7 | Improved |
| add language input (ablation) | Ours (gesture features) · YouRefIt Image ERU test (IoU=0.5, all sizes) | accuracy (%): 30.6 -> 40.5 | Improved |
| add pointing saliency heatmap to PAF features (ablation) | Ours · YouRefIt Image ERU test (IoU=0.5, all sizes) | accuracy (%): 37.6 -> 40.5 | Improved |
| add pointing saliency heatmap to PAF features (ablation) | Ours · YouRefIt Image ERU test (IoU=0.25, all sizes) | accuracy (%): 52.6 -> 54.7 | Improved |
| add PAF gestural features to ReSC-style model | ReSC · YouRefIt Image ERU test (IoU=0.5, all sizes) | accuracy (%): 34.9 -> 37.6 | Improved |
| use pretrained REF models without embodied training data | ReSC · YouRefIt Image ERU test (IoU=0.25, all sizes) | accuracy (%): 49.2 -> 20.8 | Hurt |
| add temporal module (Transformer) vs frame-based | Video ERU · YouRefIt Video ERU, ground-truth canonical frames (IoU=0.5, all sizes) | accuracy (%): 41.7 -> 38.8 | No clear effect |
| add temporal module (ConvLSTM) vs frame-based | Video ERU · YouRefIt Video ERU, canonical frame detection | average F1: 34.5 -> 45.4 | Improved |
| add temporal module (Transformer) vs frame-based | Video ERU · YouRefIt Video ERU, canonical frame detection | average F1: 34.5 -> 39.1 | Improved |
Closest papers by meaning
- Understanding Embodied Reference with Touch-Line Transformer (ICLR 2023 · similarity 0.91)
- EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents (ICML 2025 · similarity 0.87)
- Ges3ViG : Incorporating Pointing Gestures into Language-Based 3D Visual Grounding for Embodied Reference Understanding (CVPR 2025 · similarity 0.86)
- MultiPLY: A Multisensory Object-Centric Embodied Large Language Model in 3D World (CVPR 2024 · similarity 0.86)
- EmbodiedScan: A Holistic Multi-Modal 3D Perception Suite Towards Embodied AI (CVPR 2024 · similarity 0.86)
- The Language of Motion: Unifying Verbal and Non-verbal Language of 3D Human Motion (CVPR 2025 · similarity 0.86)
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.
