1. Brain Tumor Classification
Using a Kaggle dataset of MRI scans categorized into:
yes/: 155 images (with tumor)
no/: 98 images (without tumor)
Tech Stack:
Transfer Learning with ResNet50
Frameworks: TensorFlow/Keras
Deployed with Streamlit for real-time predictions & confidence display
2. Brain Tumor Segmentation
Built a U-Net model for pixel-wise segmentation of tumor regions using another Kaggle dataset:
110 folders
3,929 MRI images
3,929 segmentation masks
Tech Stack:
U-Net architecture (Functional API)
Visualization with OpenCV & matplotlib
Data handling with train_test_split (scikit-learn)
TensorFlow/Keras for training
Deployed with Streamlit
