Description
This project uses a custom-built Convolutional Neural Network (CNN) to detect human emotions from facial images in real-time. Trained on the FER-2013 dataset, it classifies emotions like Happy, Sad, Angry, Fear, Disgust, Surprise, and Neutral using 48×48 grayscale images. Ideal for beginners and intermediate AI enthusiasts, the model is lightweight (~25MB), accurate, and optimized with smart callbacks for better generalization.
– Built in TensorFlow, uses OpenCV, and ready for Kaggle GPU Notebooks.
Reviews
There are no reviews yet.