LiDAR-LLM Nu-Grounding: QA Dataset for Autonomous Driving

A question-answering dataset designed for training multimodal large language models (MLLMs) for grounding tasks in autonomous driving scenarios. It is built upon the NuScenes dataset, a known collection for self-driving research. The dataset was authored by Senqiao and last updated on the Hugging Face platform in November 2024.

Use Cases

• Train multimodal large language models (MLLMs) for visual grounding based on the described QA format.
• Benchmark model performance on autonomous driving scene understanding tasks.
• Develop models that can answer questions about objects and scenes in driving environments.
• Research the integration of language and sensor data for robotic perception.

Strengths

• Built upon the established NuScenes dataset, which provides a known foundation for autonomous driving data.
• Specifically designed for the concrete task of training MLLMs for grounding, as described in the associated paper (arXiv:2312.14074).
• Last updated on 2024-11-28, indicating recent maintenance.

Limitations

• Dataset size, row count, and file formats are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• License information is unknown, which is a critical consideration for use.

Provenance

Source

Built upon the NuScenes dataset.

Collection Method

Derived for creating a QA dataset, as described in the paper 'LiDAR-LLM: Exploring the Potential of Large Language Models for 3D LiDAR Understanding'.

Freshness

Last updated 2024-11-28 17:39:27.