ImageClassifierAPI

Project Summery

🚀 Features

• ✅ Upload labeled images via API
• ✅ Upload ZIP dataset with folder/class structure
• ✅ Automatic train/val split
• ✅ Incremental training: retrains only with new data if model already exists
• ✅ Live predictions via API or browser UI
• ✅ Auto-saves model + class labels

🧱 Project Structure

Root Directory

• main.py – FastAPI server entrypoint
• requirements.txt – Required Python packages
• mobilenet_custom.pth – Saved model weights (after training)
• class_names.txt – Text file storing class names (one per line)
• README.md – Project documentation

/app/ – Core Application Code

• __init__.py – Makes app a Python package
• api.py – Defines all FastAPI API routes
• config.py – Central configuration (paths, hyperparameters, etc.)
• data_utils.py – Functions for saving images, unpacking ZIPs, data splits
• model_utils.py – Model loading, training, prediction logic
• templates/index.html – Jinja2 HTML template for web image prediction

/data/ – Image Dataset Directory

• raw/ – Uploaded images, organized by class
• train/ – Auto-generated training images
• val/ – Auto-generated validation images

🚀 Features

• Upload labeled images or ZIP dataset
• Automatic train/validation split
• Train with existing model or from scratch
• Predict image class via API or Web UI
• View prediction confidence
• Real-time browser interface
• MobileNetV2 transfer learning (ImageNet pretrained)

🛠️ Setup

git clone https://github.com/yourusername/image_classifier_api.git
cd image_classifier_api
python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows
pip install -r requirements.txt
uvicorn main:app --reload

Output

Server Started

Add Image

Train

Result