Abstract: -: One of the big concerns worldwide is diabetic eye disease. This can cause severe deterioration of the eyes, including permanent vision loss. Early detection of eye disorders, by effective treatment, improves the risk of survival. The proposed approach is to explore the technique of machine learning to detect diabetic patients using thermographic images of an eye and to incorporate the effect of thermal variation of abnormality in the eye structure as a diagnostic imaging tool that is helpful for clinical diagnosis by ophthalmologists. Thermal images are pre-processed and then texture characteristics based on the Gray Level Cooccurrence Matrix (GLCM) from gray images are extracted and categorized using a classifier with a variety of features. The gray-level co-occurrence matrix (GLCM), also known as the gray-level spatial dependency matrix, is a statistical method of analyzing texture that considers the spatial relationship of pixels. RGB is the most commonly used color space, and we have already addressed it in past tutorials. RGB stands for red, blue and green. What the RGB model states is that three distinct images actually make up each color image. Blue image, red image, black image. Only one matrix can characterize a normal gray scale image, but a color image is actually made up of three. The HSI color model represents each color with three components: hue (H), saturation (S), intensity (I). Various matrices. Keywords: Machine learning, CNN, image processing , Feature extraction