What
An image classification model trained on the FER2013 dataset (48×48 px grayscale face images) to detect emotions. The implementation focuses on three classes: happy, neutral, and surprise. Achieved ~66% accuracy on the test set.
Why
...
Approach
...
Results
~66% accuracy on the 3-class test set.
Stack
| Layer | Choice | |---|---| | Language | Python | | Environment | Google Colab | | Dataset | FER2013 (Kaggle) | | Notebooks | Jupyter |