Research map / Point clouds / Point cloud classification
Point Cloud Quantization Through Multimodal Prompting for 3D Understanding
Hongxuan Li, Wencheng Zhu, Huiying Xu, Xinzhong Zhu, Pengfei Zhu
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 compactness loss to alignment loss (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 69.95 -> 70.01 | No clear effect |
| add separation loss alone to alignment loss (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 69.95 -> 69.19 | Hurt |
| use both compactness and separation losses together (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 69.95 -> 71.03 | Improved |
| add point cloud adapter (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 56.73 -> 71.03 | Improved |
| add learnable prompts (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 67.66 -> 71.03 | Improved |
| add quantization module (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 67.59 -> 71.03 | Improved |
| learnable prompts instead of handcrafted template (ablation) | PCQ prototypes · ScanObjectNN PB 8-shot | accuracy (%): 69.19 -> 71.03 | Improved |
| handcrafted template instead of bare class name (ablation) | PCQ prototypes · ScanObjectNN PB 8-shot | accuracy (%): 67.66 -> 69.19 | Improved |
| text embeddings as prototypes instead of cluster centroids (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 69.60 -> 71.03 | Improved |
| text embeddings as prototypes instead of trainable codebooks (ablation) | PCQ · ScanObjectNN PB 8-shot | accuracy (%): 70.06 -> 71.03 | Improved |
| PCQ on Uni3D-Ti backbone vs standard PEFT | Uni3D-Ti · ScanObjectNN PB few-shot | accuracy (%): +5.10 (1-shot) to +0.35 (8-shot) | Improved |
| apply PCQ vs PPT, trained on OBJ and tested on BG | ULIP-2 · ScanObjectNN OBJ->BG | accuracy (%): 83.2 -> 86.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PointCLIP V2 — PointCLIP V2: Prompting CLIP and GPT for Powerful 3D Open-world Learning (ICCV 2023) · better on 1 of 1 shared result
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
- UVU: Improving Multimodal Understanding via Vision-Language Unified Autoregressive Paradigm (CVPR 2026 · similarity 0.89)
- Multimodality Helps Few-shot 3D Point Cloud Semantic Segmentation (ICLR 2025 · similarity 0.88)
- Hugging Visual Prompt and Segmentation Tokens: Consistency Learning for Fine-Grained Visual Understanding in MLLMs (CVPR 2026 · similarity 0.88)
- Exploring Vision Semantic Prompt for Efficient Point Cloud Understanding (ICML 2025 · similarity 0.88)
- More Text, Less Point: Towards 3D Data-Efficient Point-Language Understanding (AAAI 2025 · similarity 0.88)
- VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks (ICLR 2025 · 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.
