Kidney disease is a social and global health concern where early detection is crucial to reducing mortality and improving treatment outcomes. Traditional diagnostic methods are time-consuming and prone to human error. To address the issue, this study proposes an efficient automated deep learning diagnostic system using medical imaging for kidney disease detection and classification. The framework integrates DenseNet121 and EfficientNetB0 for deep feature extraction, followed by SVM, Random Forest, and XGBoost classifiers combined via soft voting. The proposed system was evaluated on 12,446 CT images encompassing four kidney classes: cyst, stone, tumor, and normal. The proposed model achieved outstanding performance metrics with an accuracy of 99.24% and an F1-score of 99%. The proposed model enables early and accurate detection of kidney disease, aiding timely treatment, especially in resource-limited settings.