cjj2923 / m-sena Goto Github PK

Note: M-SENA is a research project, and is for demonstration purposes only. We give our best effort to make the platform run smoothly but we make no guarantees about the robustness of the platform. For meticulous research purposes, we recommand using the toolkits listed below, which are dedicated to reproduceable experiments and benchmarks.

M-SENA is an open-sourced platform for Multimodal Sentiment Analysis. It aims to facilitate advanced research by providing flexible toolkits, reliable benchmarks, and intuitive demonstrations.

The platform integrates two of our research toolkits, we highly recommand advanced researchers to use these toolkits:

• MMSA: A unified framework for Multimodal Sentiment Analysis, dedicated to fair and reproduceable experiments.
• MMSA-FET: A Feature Extraction Toolkit compatible with MMSA, can extract highly customized modality features.

The platform is developed with frontend and backend seperated into two repos. Follow the instructions in these repos to set up frontend and backend on your server. You can deploy frontend and backend on separate servers.

• M-SENA Frontend
• M-SENA Backend

The platform comes with an easy-to-use GUI, a demonstration video is avaliable at Youtube.

If you find our works useful, please cite our paper.

• M-SENA: An Integrated Platform for Multimodal Sentiment Analysis

@article{mao2022m,
  title={M-SENA: An Integrated Platform for Multimodal Sentiment Analysis},
  author={Mao, Huisheng and Yuan, Ziqi and Xu, Hua and Yu, Wenmeng and Liu, Yihe and Gao, Kai},
  journal={arXiv preprint arXiv:2203.12441},
  year={2022}
}

M-SENA Platform is published under the GNU GPLv3 license.