Vegetable Image Classifier

A deep learning model for classifying 6 types of Indonesian vegetables and spices:

• Daun Bawang (Scallion)
• Seledri (Celery)
• Jahe (Ginger)
• Lengkuas (Galangal)
• Kunyit (Turmeric)
• Kencur (Lesser Galangal)

Project Structure

.
├── data/
│   ├── daun_bawang/
│   ├── seledri/
│   ├── jahe/
│   ├── lengkuas/
│   ├── kunyit/
│   └── kencur/
├── src/
│   ├── data/
│   │   ├── __init__.py
│   │   └── dataset.py
│   ├── models/
│   │   ├── __init__.py
│   │   └── cnn.py
│   ├── utils/
│   │   ├── __init__.py
│   │   └── trainer.py
│   ├── __init__.py
│   ├── train.py
│   ├── inference.py
│   └── streamlit_app.py
├── requirements.txt
└── README.md

Training

To train the model:

python src/train.py

The model is trained with the following configuration:

• Batch size: 32
• Learning rate: 0.001
• Number of epochs: 30
• Training/validation split: 80/20

The trained model will be saved in the checkpoints directory.

Best model performance achieved:

• Training accuracy: 73.54%
• Validation accuracy: 75.73%

Inference

The model can be used for inference through either:

1. FastAPI endpoint (inference.py)
2. Streamlit web interface (streamlit_app.py)

Both methods load the best checkpoint and provide predictions for the 6 vegetable categories.

To start the FastAPI server:

```bash
uvicorn src.inference:app --host 0.0.0.0 --port 8000

The API will be available at http://localhost:8000.

API Endpoints

• POST /predict
  • Upload an image file to get the prediction
  • Returns the predicted vegetable class and confidence score

Model Architecture

The CNN architecture consists of:

• Convolutional layers for feature extraction
• Batch normalization for training stability
• ReLU activation functions
• Max pooling layers for spatial reduction
• Dropouts for regularization
• Fully connected layers for classification

Streamlit Web Interface

To start the Streamlit web interface:

streamlit run src/streamlit_app.py

The web interface will be available at http://localhost:8501.

Streamlit Community Cloud

You can access here https://vegetables-image-classifier.streamlit.app/