The eyes are among the most important organs in the human body. We rely on our sense of sight to see the world and comprehend what is going on around us. It can be challenging to detect and identify some eye conditions. Keratoconus is one of the rarest and most challenging to diagnose eye conditions. A progressive corneal condition called Keratoconus causes corneal scarring and thinning (Zhou et al. in Invest Ophthalmol Visual Sci 39(7), 1117–1124, 1998). As the etiology of Keratoconus is yet unclear, early detection is highly challenging. However, doctors do believe that aggressive eye rubbing and a mix of hereditary and hormonal factors may contribute to the condition. Early Keratoconus detection makes it easier for ophthalmologists to identify the eye, and the patient is better equipped to adjust to the changes. As a result, early Keratoconus detection is necessary, and we created a model for that aim. Even though there are previously many Machine Learning and Deep Learning techniques existed, which were used in the detection of Keratoconus but the model we used is more efficient and reliable than the existing algorithms. The model which is used is a Deep Learning algorithm known as VGG-19. The VGG-19, often referred to as the Visual Geometry Group, is used to identify Keratoconus. With 19 convolution layers, the deep convolution neural network architecture VGG-19 is ideal for accurate object identification and picture categorization. As opposed to VGG-16, which has 16 Convolution layers, it is an upgraded version. Due to its higher accuracy and quicker training pace, VGG-19 outperforms the currently used techniques.