This project focuses on the early and accurate detection of breast cancer using deep learning models, leveraging transfer learning and data augmentation techniques. The system is built using convolutional neural networks (CNNs), including VGG16, VGG19, and a custom architecture called OCySNet, to classify breast ultrasound images into benign, malignant, or normal categories.

To improve model performance and address data imbalance, DCGAN (Deep Convolutional Generative Adversarial Network) is integrated for realistic image augmentation. The project includes a full training and evaluation pipeline with metrics such as accuracy, precision, recall, F1-score, AUC, and visual performance graphs.

Developed in Python using TensorFlow, Keras, and Google Colab, this project showcases expertise in deep learning, medical imaging, model comparison, and hyperparameter tuning with Keras Tuner, making it a robust tool for aiding medical diagnosis.