Detecção Inteligente de Retinopatia Diabética com Redes Neurais Convolucionais Profundas

This work describes the experimental results of using Convolutional Neural Networks to detect diabetic retinopathy in medical images. Canonical architectures were considered and the best results obtained have F1-Score equal to 75,83\% with Inception architecture, highlighting the potential of using such models in this Medical Imagiology problem.