---
language:
- en
- zh
task_categories:
- spoken-language-understanding
- intent-classification
- slot-filling
---
# TeleVRSLUBench
## Dataset Description
**TeleVRSLUBench** is a spoken language understanding (SLU) benchmark that incorporates **visual scene information and explicit reasoning processes** for **joint intent detection and slot filling**.
The dataset is proposed in the paper:
> *Introducing Visual Scenes and Reasoning: A More Realistic Benchmark for Spoken Language Understanding*
It is, to the best of our knowledge, the **first SLU benchmark** that integrates scene-level visual context and reasoning information to support more realistic spoken language understanding.
The official code repository is available at:
https://github.com/Tele-AI/TeleVRSLUBench
## Citation
If you find **TeleVRSLUBench** helpful to your research, please consider citing the following papers.
We sincerely thank the original authors for their valuable contributions.
If VRSLU is useful to y…
