My diabetic retinopathy project is an end-to-end computer vision pipeline that classifies fundus images into five stages. I fine-tuned a pretrained ResNet-50, addressed class imbalance with stratified sampling and image augmentation, and used Grad-CAM to visualize model focus.
At Picsart Academy, I studied machine learning, deep learning, computer vision, and NLP, including building neural networks in PyTorch. I work with Python and tools including PyTorch, scikit-learn, OpenCV, NumPy, pandas, and SQLite.
