---
license: mit
task_categories:
- video-text-to-text
configs:
- config_name: default
  data_files:
  - split: test
    path: test.jsonl
dataset_info:
  features:
  - name: video_id
    dtype: string
  - name: metaphor_type
    sequence: string
  - name: title
    dtype: string
  - name: golden_answers
    sequence: string
  - name: frames
    sequence: string
  splits:
  - name: test
    num_examples: 861
---

# MetaphorVU: Towards Metaphorical Video Understanding

Paper: [MetaphorVU: Towards Metaphorical Video Understanding](https://huggingface.co/papers/2605.25461)

Code: [https://github.com/icip-cas/MetaphorVU](https://github.com/icip-cas/MetaphorVU)

We propose MetaphorVU-Bench, a comprehensive benchmark for metaphorical video understanding, characterized by a well-founded systematic taxonomy, metaphorical videos curated from billions of real-world candidates, and rigorous human annotation.

![](./benchmark.png "相对路径演示")

![](./construction.png "相对路径演示")