This paper presents a technique to detect exudates in fundus images, which are crucial for diagnosing diabetic retinopathy to prevent vision loss. The detection process utilizes the green channel of RGB images, applies preprocessing steps, discrete wavelet transform, and features extracted for classification through methods like KNN, SVM, and Neural Networks. If exudates are identified, the regions of interest are extracted using Canny edge detection and morphological operations to assess their severity.