OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Multimodal Sentiment Analysis

Multimodal Sentiment Analysis is an open-source Python research project implementing attention-based fusion for sentiment classification in user-generated videos and conversational data. It provides code associated with research papers and documents experiments using MOSI, MOSEI and IEMOCAP, with options for attention-based or concatenation-based fusion and unimodal or multimodal training.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Multimodal Sentiment Analysis — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it as a research codebase when investigating multimodal sentiment experiments or adapting the documented dataset and fusion configurations.

Who it is for

  • Researchers studying multimodal sentiment or emotion classification.
  • Developers exploring attention-based fusion of audio, text and visual features.
  • Students reproducing or adapting experiments based on the cited research.

Documented uses

  • Explore sentiment classification across audio, text and video modalities.
  • Compare attention-based fusion with concatenation-based fusion.
  • Configure experiments using MOSI, MOSEI or IEMOCAP and their documented label counts.
  • Study utterance-level attention and speaker-independent train/test data organization.
Before you choose

Requirements & limits

The repository is public and carries an MIT license. Its README says it is no longer maintained and points to a centralized repository for ongoing multimodal work.

  • The README marks the create_data.py preprocessing script as obsolete and says to use the preprocessed datasets included in the repository.
  • The documented dataset configurations use different label counts: MOSI has 2, MOSEI 3 and IEMOCAP 6.
  • Fusion can be set to attention-based or concatenation-based; the README notes that --fusion True applies to the multimodal network.
  • The repository's maintainers say it is no longer maintained.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run a multimodal fusion configuration

Use a provided dataset configuration to launch a multimodal experiment with attention-based fusion.

Before you start

  • The repository's Python code and preprocessed dataset files.
  • Choose one documented dataset: MOSI, MOSEI or IEMOCAP, with its corresponding class count: 2, 3 or 6.

Steps

  1. Select the dataset and matching number of classes.

  2. Run the command below, substituting the dataset and class count as appropriate.

    Documented command
    python run.py --unimodal False --fusion True --data mosei --classes 3

Expected outcome

Starts the model with multimodal fusion enabled and the selected dataset configuration.

Guide limitations

  • The repository is no longer maintained.
  • The README documents sample commands but does not provide environment installation or dependency setup instructions in the supplied material.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
soujanyaporia/multimodal-sentiment-analysis · Official homepage title · retained in its source language
How we review sourcesSuggest a correction