A 5-layer CNN model for facial emotion recognition trained on FER-2013. Achieved 76% validation and 63% test accuracy using data augmentation. Key features include convolutional layers, max pooling, and dropout. Suitable for human-computer interaction applications.
python machine-learning deep-learning tensorflow numpy scikit-learn keras cnn pandas seaborn classification matplotlib regularization augmentation tf early-stopping efficientnetb4 fer-2013 model-chackpoint
-
Updated
Apr 12, 2025 - Jupyter Notebook