Identification and classification of brain tumors have become a concern for the medical community as the treatment planning and patients’ prognosis is highly dependent on their classification. In the past few years, deep learning models have become the most discussed and widely used tools to achieve higher accuracy and efficiency in this area. This study gives a comparison of the two widely used deep learning networks, the modified UNet and ResNet-50 for the task of brain tumor classification. The study assesses the efficiency of these architectures on the benchmark dataset and indicates the strengths and limitations of each. The evaluation process is made of extremely rigorous testing and validation procedures, which are aimed at making the data reliable. The outcomes of our study indicate that both modified UNet and ResNet-50 can reasonably be applied to the brain tumor classification. Modified UNet surpasses ResNet-50 in overall accuracy, which implies its effectiveness in high-precision classifications. This comparison revealed the feasibility of the architectures for classifying brain tumors. It helps both researchers and practitioners in the field by providing them with information to find the model appropriate for their specific needs. Moreover, the results of the study could be used to build the discussion on the application of the deep learning models in the diagnosis and classification of brain tumors as a part of the medical image analysis. Therefore, this paper provides a thorough guidance for individuals who want to use deep learning to enhance healthcare outcomes.