Research map / Vision-language and multimodal / Human and embodied modeling
HAMMER: Harnessing MLLMs via Cross-Modal Integration for Intention-Driven 3D Affordance Grounding
Lei Yao, Yong Chen, Yuejiao Su, Yi Wang, Moyun Liu, Lap-Pui Chau
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 object category to the prompt and predict affordance label as auxiliary output (ablation) | HAMMER · PIAD Unseen | aIOU (%): 8.78 -> 13.71 | Improved |
| add object category to the prompt (ablation) | HAMMER · PIAD Unseen | aIOU (%): 12.35 -> 13.71 | Improved |
| add hierarchical cross-modal integration (ablation) | HAMMER · PIAD Unseen | aIOU (%): 10.50 -> 13.71 | Improved |
| add multi-granular geometry lifting (ablation) | HAMMER · PIAD Unseen | aIOU (%): 10.20 -> 13.71 | Improved |
| add both integration and geometry lifting (ablation) | HAMMER · PIAD Seen | aIOU (%): 19.55 -> 22.20 | Improved |
| add textual language modeling loss (ablation) | HAMMER · PIAD Unseen | aIOU (%): 9.78 -> 13.71 | Improved |
| use multi-granular lifting instead of single-stage lifting (ablation) | HAMMER · PIAD Unseen | aIOU (%): 12.26 -> 13.71 | Improved |
| use multi-granular lifting instead of concatenation lifting (ablation) | HAMMER · PIAD Unseen | aIOU (%): 11.45 -> 13.71 | Improved |
| fine-tune MLLM language component with LoRA instead of freezing (ablation) | HAMMER · PIAD Unseen | aIOU (%): 7.55 -> 13.71 | Improved |
| add Stage I (bottleneck cross-attention) of hierarchical integration (ablation) | HAMMER · PIAD Unseen | aIOU (%): 11.85 -> 13.71 | Improved |
| swap PointNet++ (from scratch) for pretrained Uni3D backbone (ablation) | HAMMER · PIAD Seen | aIOU (%): 22.20 -> 24.67 (seen); 13.71 -> 12.28 (unseen) | Mixed |
| use Qwen2.5-VL as MLLM instead of segmentation-tuned LISA / LISA++ (ablation) | HAMMER · PIAD Unseen | aIOU | Improved |
Later papers that compared with this method
3 later accepted papers report HAMMER as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- The Coherence Trap: When MLLM-Crafted Narratives Exploit Manipulated Visual Contexts (CVPR 2026 · better on 10 of 10 shared results)
- Bridging Pixels and Words: Mask-Aware Local Semantic Fusion for Multimodal Media Verification (CVPR 2026 · better on 4 of 4 shared results)
- OmniVL-Guard: Towards Unified Vision-Language Forgery Detection and Grounding via Balanced RL (ICML 2026 · better on 3 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GREAT: Geometry-Intention Collaborative Inference for Open-Vocabulary 3D Object Affordance Grounding (CVPR 2025 · similarity 0.92)
- SeqAfford: Sequential 3D Affordance Reasoning via Multimodal Large Language Model (CVPR 2025 · similarity 0.91)
- Grounding 3D Object Affordance with Language Instructions, Visual Observations and Interactions (CVPR 2025 · similarity 0.91)
- AffordBot: 3D Fine-grained Embodied Reasoning via Multimodal Large Language Models (NeurIPS 2025 · similarity 0.90)
- Learning 2D Invariant Affordance Knowledge for 3D Affordance Grounding (AAAI 2025 · similarity 0.90)
- IAAO: Interactive Affordance Learning for Articulated Objects in 3D Environments (CVPR 2025 · similarity 0.89)
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.
