Purpose: Early detection of eye disease from fundus images is an important factor in eye care and preventing vision loss. Methods: Deep learning models are used nowadays to classify eye diseases like glaucoma and diabetic retinopathy. We propose a deep learning model with four convolutional blocks, followed by a spatial attention block and three fully connected blocks. Results: We evaluated the model’s performance using Precision, Recall, and F1-score across five categories: diabetic retinopathy, glaucoma, maculopathy, optic atrophy, and normal fundus images. The model achieved an overall classification accuracy of 98%. Conclusion: The model accurately classified fundus images into four disease classes and a control class.