A fully offline image classification api built with FastAPI (backend) and Streamlit (frontend).
Uses pre-trained PyTorch models (ResNet50) to classify images and display the top 5 predictions.
This project works completely offline once dependencies are installed, requiring no internet connection.
- ✅ Fully offline and free to use
- ✅ CPU-friendly (no GPU required for small models)
- ✅ Upload images through a simple Streamlit interface
- ✅ Returns top 5 ImageNet predictions with probabilities
- ✅ Easily switch to other PyTorch pre-trained models (e.g., EfficientNet)