3DFroMLLM: 3D Prototype Generation only from Pretrained Multimodal LLMs
This work addresses the challenge of enhancing 3D spatial understanding in AI systems for applications like computer vision and robotics, representing a novel method for a known bottleneck rather than a foundational advancement.
The paper tackles the problem of limited spatial reasoning in multimodal LLMs by introducing 3DFroMLLM, a framework that generates 3D object prototypes with geometry and part labels directly from pretrained models, resulting in a 15% improvement in image classification pretraining and a 55% accuracy boost for part segmentation in fine-grained vision-language models.
Recent Multi-Modal Large Language Models (MLLMs) have demonstrated strong capabilities in learning joint representations from text and images. However, their spatial reasoning remains limited. We introduce 3DFroMLLM, a novel framework that enables the generation of 3D object prototypes directly from MLLMs, including geometry and part labels. Our pipeline is agentic, comprising a designer, coder, and visual inspector operating in a refinement loop. Notably, our approach requires no additional training data or detailed user instructions. Building on prior work in 2D generation, we demonstrate that rendered images produced by our framework can be effectively used for image classification pretraining tasks and outperforms previous methods by 15%. As a compelling real-world use case, we show that the generated prototypes can be leveraged to improve fine-grained vision-language models by using the rendered, part-labeled prototypes to fine-tune CLIP for part segmentation and achieving a 55% accuracy improvement without relying on any additional human-labeled data.