Vision impairment and blindness are chronic diseases where blindness is a complete or partial loss of vision. Blindness occurs suddenly or over a period. The primary reasons for blindness occurrence are diabetes and secondly eye diseases. Older people in developing countries are more affected than other age people. The major problem with blindness is no proper guidelines or precautions for the people. According to the World Health Organization (WHO), 2.2 Billion people are suffering from near or distance impairment. Due to age, the leading causes are uncorrected refractive errors and cataracts. Diabetes patients mostly face vision problems due to diabetic retinopathy. The high blood glucose level in the eye blood vessels increases the chances of vision problems. We proposed the model using deep learning algorithms to detect blindness in the early stages. We apply pre-processing approaches to manage the dataset. Then apply ResNet, DenseNet, Xception, and InceptionResNet models to train the model. The trained model was used for the testing and evaluate the proposed model using accuracy, precision, recall, and f1-score. The proposed model outperformed using the ResNet model compared to the other models. This model can be utilized for clinical purposes after testing on different datasets. The proposed model evaluated for accuracy, precision, recall, and f-measure are 0.93, 0.94, 0.98, and 0.94 respectively. The results show proposed model outperforms blindness detection.