<p>Cervical cancer remains a significant global health challenge, particularly due to the complexities involved in accurately detecting and classifying various stages of the disease. The early detection and differentiation between precancerous and cancerous lesions are critical for effective treatment and improved patient outcomes. Traditional diagnostic methods, including cytology and histology, often suffer from limitations such as subjectivity and variability in results. To address these challenges, this study proposes a novel approach utilizing a custom multiscale residual network, complemented by Swin Transformer-based preprocessing of cervical images. The proposed method leverages advanced deep learning techniques to enhance the classification accuracy of cervical images into healthy, precancerous, and cancerous categories. The model’s performance is evaluated against a set of established metrics, including accuracy, precision, sensitivity, specificity, F1-score, and Matthews Correlation Coefficient (MCC), demonstrating significant improvements in classification accuracy. By integrating Swin Transformer preprocessing with a custom multiscale residual network, this study achieves a classification accuracy of 98.93%, showcasing the potential of this approach to provide a robust, automated solution for cervical cancer detection. The results underscore the effectiveness of combining advanced image preprocessing with deep learning models to address the complexities of cervical cancer classification and offer a promising direction for future research in this critical area.</p>