Image Classification with Transfer Learning

This project utilizes transfer learning to classify images into categories using a pre-trained VGG16 model. Transfer learning allows us to leverage the features learned by the VGG16 model on the ImageNet dataset to improve classification performance on a new dataset.

Project Structure

• data/: Contains scripts for loading and preprocessing image data.
• model/: Contains the transfer learning model definition.
• scripts/: Contains scripts for training and evaluating the model.
• requirements.txt: Lists the required Python packages.

Getting Started

Open Colab file

OR

1. Clone the repository:

   git clone https://github.com/coderooz/image_classification_transfer_learning.git
   cd image_classification_transfer_learning

2. Install dependencies:

   pip install -r requirements.txt

3. Prepare your dataset: Place your image data in data/train/ and data/validation/ directories, with subdirectories for each class.

4. Train the model:

   python scripts/train_model.py

5. Evaluate the model:

   python scripts/evaluate_model.py

Results

The model's accuracy on the validation set will be printed after evaluation.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements

• The VGG16 model is provided by TensorFlow.

Contact

• Ranit Saha - Coderooz