Early diagnosis and patient treatment are greatly aided by the early detection of brain tumors. In this study, we give a thorough investigation on YOLOv8 model-based deep learning-based brain tumor identification. To train and test our model, we used the BRATS2023 dataset, which consists of high-resolution 3D brain MRI Images. We used pre-trained weights from a large-scale dataset, such as the COCO dataset, to initialize the YOLOv8 architecture, which significantly improved the performance of our model. This initialization made it possible for our model to use information gleaned from a variety of objects, improving the detection accuracy of brain tumors. Additionally, it made it easier to find tumors of varied sizes and forms, strengthening and expanding the model. Our research’s focus on increasing training speed without sacrificing accuracy is one of its major achievements. We significantly cut training time while retaining good detection accuracy by using optimization approaches including batch normalization and sophisticated data augmentation methodologies. This development is especially helpful in the medical industry, where prompt and precise diagnosis are crucial.