Cervical cancer (CC) is the fourth most frequent form of cancer in women. The HPV-related condition most usually associated with HPV infection is CC in general. One of the often-used methods to find CC is the pap smear. However, some drawbacks include a long wait for findings, a reduction in sensitivity, and a reduction in laboratory quality control. In this research, a novel CER-XNET approach to identifying CC and reducing death rates has been proposed to address these drawbacks. Images from the dataset are initially acquired for MRI cervical scans. The classification of several MRI scans of CC is then done utilizing the deep learning (DL) network. Next, feature extraction is performed on the pre-processed images using capsule network. Using an Xception network, the images are then divided into normal and abnormal classes. The suggested based on its F1score, precision, recall, specificity, and accuracy. The accuracy of the suggested CER-XNET method is 99.24%. The accuracy of the L1reg CNN network strategy outperformed the majority of the reported existing methods, which suggests successful outcomes.