The primary objective of this project is to develop an automated system for Diabetic Retinopathy (DR) detection that is both accurate and scalable. The key objectives include: 1. Data Preprocessing: To clean and preprocess the Diabetic Retinopathy Detection dataset, ensuring the quality of the input images and addressing issues such as noise and image variability. 2. Model Development: To implement and train deep learning models such as MobileNet, ResNet50, and hybrid models combining MobileNet + GNN and MobileNet + RNN. These models will classify DR into five stages, improving accuracy by learning both spatial and temporal patterns in retinal images. 3. Active Learning (AL): To integrate Active Learning (AL) into the model training process, enabling the system to actively select and learn from the most informative samples, reducing the reliance on large labeled datasets and improving model efficiency. 4. Model Evaluation: To evaluate the performance of the models based on key metrics such as accuracy, precision, recall, and F1-score. Additionally, the models will be assessed for their ability to generalize to new, unseen data. 5. Scalability: To ensure that the developed system can be scaled to work with larger datasets and deployed in low-resource environments, providing an efficient and accessible solution for DR detection.