ControlNet / emolysis

[ACII Demo] Emolysis: A Multimodal Open-Source Group Emotion Analysis and Visualization Toolkit

Home Page:https://ieeexplore.ieee.org/abstract/document/10970223

Repository from Github https://github.comControlNet/emolysisRepository from Github https://github.comControlNet/emolysis

Emolysis: A Multimodal Open-Source Group Emotion Analysis and Visualization Toolkit

This repo is official repository for the paper Emolysis: A Multimodal Open-Source Group Emotion Analysis and Visualization Toolkit.

Get Started

We provided a static demo review for you to try. Please visit https://emolysis.controlnet.space/local/1.

If you want to analyze your own video, please follow the instructions below to deploy the server.

Deploy the Server

From Docker (x86 with CUDA)

Requires:

  • Docker
  • nvidia-docker

Run the server.

docker run --runtime nvidia -p <PORT>:8000 [-v <CACHE_DIR_FOR_MODELS>:/app/checkpoints] --name emolysis controlnet/emolysis

Then, you can access the app at http://127.0.0.1:<PORT>.

From Source

Requires:

  • Conda
  • Node.js

Install dependencies.

npm install
npm run build
cd service
bash -i build_env.sh  # use `build_env.mac.sh` for arm-based mac
conda activate emolysis
cd ..

Run the server.

python service/main.py --port <PORT>

Then, you can access the app at http://127.0.0.1:<PORT>.

References

If you find this work useful for your research, please consider citing it.

@inproceedings{ghosh2024emolysis,
  title={Emolysis: A multimodal open-source group emotion analysis and visualization toolkit},
  author={Ghosh, Shreya and Cai, Zhixi and Gupta, Parul and Sharma, Garima and Dhall, Abhinav and Hayat, Munawar and Gedeon, Tom},
  booktitle={2024 12th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW)},
  pages={116--118},
  year={2024},
  organization={IEEE}
}

About

[ACII Demo] Emolysis: A Multimodal Open-Source Group Emotion Analysis and Visualization Toolkit

https://ieeexplore.ieee.org/abstract/document/10970223

License:Apache License 2.0


Languages

Language:Vue 46.6%Language:Python 34.5%Language:TypeScript 13.6%Language:HTML 2.9%Language:Shell 1.1%Language:Dockerfile 0.8%Language:CSS 0.4%Language:JavaScript 0.2%