Detecting mitotic tumor cells is crucial for accurate breast cancer diagnosis and treatment. Currently, pathologists perform this task manually, which is time-consuming, error- prone, and subjective. However, advancements in scanning technologies now enable us to capture high-resolution images of biopsy slides. Leveraging these advancements, we can develop deep learning algorithms to assist pathologists in automating mitotic tumor cell detection in histopathological images. This holds the promise of enhancing the efficiency and accuracy of breast cancer diagnosis and treatment. In this study, we employed the Detectron2 framework, utilizing the FasterRCNN algorithm with a Resnet50 with Feature Pyramid Network (FPN) backbone to identify mitotic cells in histopatologic images of breast cancer. Additionally, we explored the effects of color normalization and color transfer techniques on mitotic cell detection. By experimenting with different hyperparameter combinations and optimizers, we aim to compare and analyze the effectiveness of these approaches. The best-performing model demonstrated a precision of 0.6 and a mean average recall of 0.47.