README.md

---
title: OneEmo Affective Reasoning
emoji: 🎭
colorFrom: gray
colorTo: pink
sdk: gradio
sdk_version: 6.22.0
app_file: app.py
short_description: Unified multimodal reasoning for emotion perception
python_version: "3.12"
startup_duration_timeout: 30m
---

# OneEmo: A Unified Multimodal Reasoning Model for Emotion Perception

This Space demonstrates [OneEmo](https://huggingface.co/Jiaha0Hu4ng/OneEmo), a unified
multimodal reasoning model for emotion perception, understanding, and interaction in
video scenarios. It supports eight tasks:

- **B-MER** – Basic Multimodal Emotion Recognition (single label)
- **MSA** – Multimodal Sentiment Analysis (positive/neutral/negative)
- **OV-MER** – Open-Vocabulary Multimodal Emotion Recognition
- **MHD** – Multimodal Humor Detection
- **MSD** – Multimodal Sarcasm Detection
- **ERG** – Empathetic Response Generation

Upload a video, select a task, and click **Analyze** to see the model's emotion
analysis with chain-of-thought reasoning.

## References

- Paper: [OneEmo: A Unified Multimodal Reasoning Model for Emotion Perception, Understanding, and Interaction](https://arxiv.org/abs/2608.06013)
- Model: [Jiaha0Hu4ng/OneEmo](https://huggingface.co/Jiaha0Hu4ng/OneEmo)
- Code: [GitHub](https://github.com/waHAHJIAHAO/OneEmo)

Example videos are CC0-licensed, sourced from the
[Gradio Space Example Inputs — Videos](https://huggingface.co/datasets/linoyts/repo-to-space-example-videos) dataset.