Diabetic retinopathy is a major complication of diabetes that can lead to vision loss. Early detection and diagnosis are crucial for effective management. This study proposes a ResNet-101 based deep learning model for automated diabetic retinopathy detection. The model is trained on a dataset of retinal fundus images. ResNet-101 architecture is utilized for its ability to learn complex features. The model is fine-tuned to classify different stages of diabetic retinopathy. Experimental results demonstrate high accuracy and sensitivity. The proposed approach has the potential to assist clinicians in early detection. This can lead to timely interventions and improved patient outcomes. The model can be integrated into clinical workflows for efficient diagnosis. Future work includes collecting larger datasets and exploring multi-modal fusion. The study highlights the effectiveness of deep learning in diabetic retinopathy detection. ResNet-101 shows promise in analyzing retinal images. This approach can reduce the burden on clinicians and improve diagnosis accuracy. Diabetic retinopathy detection using ResNet-101 is a significant step towards automated diagnosis. The proposed model can be extended to detect other retinal diseases. Overall, the study demonstrates the potential of ResNet-101 in improving diabetic retinopathy diagnosis. The results are encouraging and warrant further research. The study contributes to the growing body of research on deep learning in medical imaging. By leveraging ResNet-101, this approach can improve patient care and outcomes.