Practical overview
When it fits
Use the repository when you need research implementations and are prepared to select a specific model, satisfy its dependencies, and work with the corresponding dataset or prepared features.
Who it is for
- Researchers comparing conversational emotion-recognition architectures.
- Developers prototyping dialogue emotion analysis using research code.
- Teams studying conversational context or emotion causes with prepared datasets.
Documented uses
- Compare implementations of emotion recognition in conversation.
- Train a documented DialogueGCN configuration on IEMOCAP.
- Explore models that incorporate speaker relationships and dialogue context.
- Review multimodal emotion-detection approaches involving text, audio, visual signals, or conversational video.
Requirements & limits
The repository is public and specifies an MIT license.
- Requirements differ by model: the repository includes PyTorch, TensorFlow, and Keras implementations.
- The README lists older, implementation-specific dependency versions; for example, its two DialogueGCN paths specify different PyTorch Geometric versions.
- Some workflows require prepared dataset features; DialogueRNN lists IEMOCAP and AVEC training scripts.
- The README directs readers seeking visual and audio feature extraction guidance to MUStARD.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Train the listed DialogueGCN IEMOCAP configuration on CPU
Run the README's specified IEMOCAP DialogueGCN training command without CUDA.
Before you start
- Python 3
- PyTorch 1.0
- PyTorch Geometric 1.3
- Pandas 0.23
- Scikit-Learn 0.20
Steps
Use the repository's DialogueGCN instructions and prepare the IEMOCAP dataset for that implementation.
Run the command below.
python train_IEMOCAP.py --base-model 'LSTM' --graph-model --nodal-attention --dropout 0.4 --lr 0.0003 --batch-size 32 --class-weight --l2 0.0 --no-cuda
Expected outcome
The command runs the documented IEMOCAP training configuration on CPU; the README reports weighted average F1 results from its own runs.
Guide limitations
- The README warns that PyTorch Geometric CUDA atomic operations can introduce non-determinism and recommends this CPU command for reproducing reported results.
- The result figures are repository-reported, not an independent execution check.
Preprocess and train the DialogueGCN-mianzhang IEMOCAP implementation
Follow the README's two script stages for the IEMOCAP corpus.
Before you start
- Python 3
- PyTorch 1.0
- PyTorch Geometric 1.4.3
- Pandas 0.23
- Scikit-Learn 0.20
Steps
Run
./scripts/iemocap.sh preprocess.Run
./scripts/iemocap.sh train.
Expected outcome
The documented preprocessing and training stages are invoked for the IEMOCAP corpus.
Guide limitations
- The README lists different PyTorch Geometric requirements for this implementation than for the other DialogueGCN instructions.
- The README's reported comparison is for IEMOCAP and should not be treated as a result for other datasets.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 16 October 2026
- Original website title
- declare-lab/conv-emotion · Official homepage title · retained in its source language