Cervical cancer is a severe medical issue around the world and the primary reason for mortality in females. The study introduces a hybrid framework that incorporates transfer learning along with the DenseNet121 to perform feature extraction, PCA to reduce the dimensionality, and ensemble classification by using SVM and RF to increase detection accuracy. Class imbalance is addressed by using SMOTE for oversampling the minority class while balancing efficiency computation with PCA, and for achieving superior performance on Mendeley LBC data, the framework produced 98.78% accuracy and F1 score, recall, and precision of 98%. Our study uses a hybrid approach that combines deep learning with traditional machine learning models for cervical cancer detection. The proposed framework is scalable and efficient, has high accuracy, and is best suited for deployment in resource-constrained environments.