Background <p>Cervical cancer is a largely preventable disease that affects more than half a million women worldwide, with approximately 90% of the occurrences being in low- and middle-income countries. Regular screening for changes preceding cancer can reduce occurrences by enabling early preventative treatment of the affected individuals. The use of low-cost, mobile-based applications can serve as an aid to physicians, countering issues, such as difficulty obtaining high-end instruments, limitations in resource availability and expert personnel as well as variability in human judgment.</p> Methods <p>Twenty-five papers relevant to the above problem for the past five years (2019–2024) are surveyed to identify the current solutions and potential areas of research.</p> Results <p>It was seen that most current literature uses highly performant but heavyweight CNN architectures for image classification. A notable difficulty in evaluating the relevant performance is the absence of either a standard dataset or a standard classification policy, disallowing the direct comparison of classification parameters. It was noted that transfer learning from MobileNet, YOLO and ResNet architectures, combined with image segmentation techniques, showed good results. Techniques making use of sequential images or a combination of VIA and VILI images show promise, as well as those combining architectures into ensembles, or integrating them into novel hybrids.</p> Conclusion <p>Mobile-based screening has high potential; but, standardisation and replicability in the current landscape needs improvement.</p>