Many perspectives have been grown and extended instantaneously due to the evolution of the Fourth Industrial Revolution. Brain tumor detection is one of the most crucial mechanisms for standardization and care for injured patients. Early diagnosis from the beginning state lets the medical team develop comprehensive recovery protocols that help enhance patients’ survival rates. We have deployed the k-means clustering algorithm to stratify samples into three different view angles of MRI images (transverse, coronal, and sagittal) and combined a modified Residual Network (ResNet) architecture to diagnose three brain tumor types: glioma and meningioma pituitary tumor and recognize MRI images without tumor. The approach is evaluated on the dataset from Nanfang Hospital and General Hospital, Tianjin Medical University, China, with MRI images. Our result achieved 96% in brain tumor classification accuracy, the best among considered famous pre-trained networks.