🧠 The CNN Model
Our CNN model built with PyTorch achieves up to 90% accuracy on the dataset. We used ReLU activations, dropout, and adaptive learning rates for optimal performance.
A deep learning model that decodes human emotion from facial images — powered by CNNs & crafted with care ❤️
Try Live DemoFaceSense is an AI-driven facial expression recognition system trained on 7 emotions using CNN architecture. It includes a well-curated dataset, a handcrafted model, and a futuristic frontend UI.
Seven classes: Angry, Disgust, Fear, Happy, Neutral, Sad, Surprise — organized and ready for deep learning pipelines. All images were manually labeled and organized using Python scripts.
Our CNN model built with PyTorch achieves up to 90% accuracy on the dataset. We used ReLU activations, dropout, and adaptive learning rates for optimal performance.
Experience real-time facial emotion recognition using your webcam. Works best on desktop with camera permissions enabled.
Start Demo
Lead Developer
Model Trainer & Developer
Deployment
QA & Docs
Explore the complete codebase, dataset and notebook.
GitHub Repo