Kidney stone is a crystal concretion in impacts human health and affects the kidney which is a prevalent health issue associated with our modern lifestyle. It is a growing urological condition that negatively 10% of the global population. Threating the disease is according to the stone characteristics. Patients can properly cure from the diseases if they gate proper treatments according to the stone grade. The current grading of stone stage is done by on the domain of urologists. Different stone stage threated with different medication and other way of controlling technique. While various studies have focused on kidney disease identification and detection, there is a dearth of research specifically targeting kidney stone grading on the current technology. This study we focused on kidney stone stage classification using computer vision and machine learning techniques. To achieve the objective of the studies Computed Tomography (CT) images are collected from Tibeba Gihon referral hospital, GAMBY Teaching General hospital, and Felege Hiwot referral hospital. The collected dataset was prepared with preprocessing techniques like noise removal, image enhancement, and segmentation. For selecting the best noise removal technique image quality evaluation metrics are used. Also, we can apply Feature extraction techniques using Histogram Oriented Gradient (HOG), Morphological, and VGG16 for SVM, CNN, and KNN classifiers. From the three classifiers despite the challenge of detecting class four and class five instances SVM model with VGG16 feature outperform better result with 95% accuracy. Finally, we have a Testing accuracy of 93 % in SoftMax using CNN. We can use the Python programming tool (Jupyter Notebook) for implementation with a tenser flow and the supported libraries will be applied. Adding more dataset and health status as well as used better feature for detection of similar appearance in different class is suggested for future researchers. Keywords: Deep ensemble learning, feature extraction, Generative Adversarial Network, Gray Level Co-occurrence matrix, Histogram of Oriented Gradients