How sifatahammed/analysis-of-deep-learning-models-for-emotion-recognition works
This repository is an experimental emotion-recognition project: SEED-V data is prepared and segmented, then evaluated with DNN, LSTM, CNN, and CNN-LSTM models under zero- or average-padding variants. The experiment notebooks use K-fold evaluation, and the README summarizes model and padding performance. The notebook contents were not sampled, so the graph follows the documented workflow and notebook names; it does not infer implementation-level calls or data wiring.
Main entry points: Data_Preprocessing_and_Segmentation.ipynb Preprocessing and Segmentation · README.md Reported Findings · 1. Zero padding Kfold DNN.ipynb DNN Zero Padding · 2. Average Pooling Kfold DNN.ipynb DNN Average Padding
Components 10
Data Preparation
- Preprocessing and Segmentation
Data_Preprocessing_and_Segmentation.ipynb
Model Experiments
- DNN Zero Padding
Experiments/1. Zero padding Kfold DNN.ipynb - DNN Average Padding
Experiments/2. Average Pooling Kfold DNN.ipynb - LSTM Zero Padding
Experiments/3. Zero Padding Kfold LSTM.ipynb - LSTM Average Padding
Experiments/4. Average Pooling Kfold LSTM.ipynb - CNN Zero Padding
Experiments/5. Zero padding Kfold CNN.ipynb - CNN Average Padding
Experiments/6. Average pooling Kfold CNN.ipynb - CNN-LSTM Average Padding
Experiments/7. Average Pooling Hybrid CNN-LSTM.ipynb
Evaluation Results
- Reported Findings
README.md
Other components
- SEED-V Dataset
Connections 15
- SEED-V Dataset → Preprocessing and Segmentation: provides data
- Preprocessing and Segmentation → DNN Zero Padding: prepares data
- Preprocessing and Segmentation → DNN Average Padding: prepares data
- Preprocessing and Segmentation → LSTM Zero Padding: prepares data
- Preprocessing and Segmentation → LSTM Average Padding: prepares data
- Preprocessing and Segmentation → CNN Zero Padding: prepares data
- Preprocessing and Segmentation → CNN Average Padding: prepares data
- Preprocessing and Segmentation → CNN-LSTM Average Padding: prepares data
- DNN Zero Padding → Reported Findings: reports metrics
- DNN Average Padding → Reported Findings: reports metrics
- LSTM Zero Padding → Reported Findings: reports metrics
- LSTM Average Padding → Reported Findings: reports metrics
- CNN Zero Padding → Reported Findings: reports metrics
- CNN Average Padding → Reported Findings: reports metrics
- CNN-LSTM Average Padding → Reported Findings: reports metrics