Brain tumor classification with VGG16 and GradCAM++

Try at : https://btumor-vgg.streamlit.app/

Overview

The project aims at accurately identifying the presence of brain tumors in MRI images. The project utilizes deep learning techniques, specifically the VGG16 convolutional neural network architecture, for image classification. To enhance interpretability and localize areas of interest in the MRI scans, GradCAM++ (Gradient-weighted Class Activation Mapping) is employed, providing visual explanations of the model's decisions.

Features

• Visualization with GradCAM: This feature enhances interpretability by using GradCAM (Gradient-weighted Class Activation Mapping) to generate visual heatmaps. These heatmaps highlight the regions in MRI images that are critical for the model's tumor classification.

• Multiple Image Processing: Users can upload several images simultaneously, allowing for quick evaluation of multiple scans in one go.

Classification report and Confusion matrix

Screenshots