Brain tumors pose significant challenges in the field of medical image processing. Tumor that shows no symptoms at all in its early stages is referred to as asymptomatic tumor. Therefore, the implementation of an automated method to detect brain tumor in MRI at an early stage will aid radiologists in their detection process. Deep learning models are gaining a lot of attention due to their quick advancement in MRI imaging, especially in recognizing the presence of tumor. However, if the MRI scans are complex, the accuracy of disease prediction is significantly decreased. Existing approaches, on the other hand, have limits in terms of scalability and performance, which highlights the need to conduct additional research. To overcome this problem we have proposed a comprehensive UNet3+ model that performs the detection, segmentation, and classification of tumor in a single step. The performance of the proposed model is assessed by evaluating its ability to accurately segment and classify tumor using publicly available datasets BRATs. The proposed model achieves an accuracy of 99.37%, and a F1-score of 99.64%. Based on the findings, the suggested model demonstrates superior performance compared to the existing techniques.