This paper explores two different paradigms in evolving AI technologies, those being Vision Transformers(ViT) and YOLOv8, in the context of breast cancer detection of mammography images. Although Convolutional Neural Networks (CNNs) have seen widespread usage in the field of Computer Aided Diagnosis (CAD), whilst the ViT models being compared with YOLOv8 models has largely gone unexplored. To gather results and analyze them we utilize several python libraries such as Ultralytics and Transformer to either train or fine-tune our models, we employ a dataset with a validation split of 20% and 80% training split. The results are gathered through a confusion matrix and results graphs demonstrating the models loss over epochs. This paper fills a gap in current research by directly comparing YOLOv8 and ViT’s, offering insights for future CAD systems in breast cancer detection. This thesis find that the Vision transformer outperform YOLOv8 in terms of overall accuracy, with a score 84.0% and the YOLOv8 model had an overall accuracy of 82.7%, but that the YOLOv8 has more consistent performance across the three different classes.