Image Classification Project

CNN-based image classification with FastAPI backend and Streamlit dashboard.

Screenshots

Quick Start

Easiest - GitHub Container Registry (any OS):

# Setup (one time)
./setup.sh

# Deploy (for all users)
./deploy.sh run

💡 For users: Just run ./deploy.sh run after setup
For developers: Run ./deploy.sh build then ./deploy.sh run

Local Development (no Docker):

pip install -r requirements.txt
uvicorn api.main:app --port 8000 &
API_URL=http://localhost:8000 streamlit run streamlit/app.py

Docker Build (developers):

HOST_UID=$(id -u) HOST_GID=$(id -g) docker compose up --build

Dataset

Intel Image Classification from Kaggle (~25,000 images, 6 classes)

Download: kaggle.com/datasets/puneet6060/intel-image-classification

Models

Model	Accuracy	Time
Baseline	73%	~5 min
Regularized	79%	~8 min
Transfer Learning	88%	~15 min

Documentation

• DEPLOY.md - Complete deployment guide
• RUN.md - Detailed run instructions
• DATASET_SETUP.md - Dataset configuration
• TROUBLESHOOTING.md - Common issues

Technologies

TensorFlow, FastAPI, Streamlit, Docker