MobileNet V1 Implementation from scratch

This repository contains the implementation of a MobileNet model from scratch to classify images of cats and dogs using TensorFlow and Keras.

Repository Structure

• model.py: Defines the MobileNet model architecture.
• train.py: Handles training and evaluation of the model.
• test.py: Loads an image, preprocesses it, and uses the trained model to make predictions.
• utils.py: Contains utility functions such as image preprocessing.
• main.py: The main script to run the training and testing.

Setup Instructions

1. Clone the repository:

   git clone https://github.com/smhatefi/mobilenet_v1.git
   cd mobilenet_v1
   

2. Install the required packages:

   Make sure you have Python and pip installed. Then, install the required Python packages:

   pip install tensorflow keras matplotlib tensorflow-datasets
   

3. Download sample images for testing:

   The main.py script downloads sample images for testing.

How to Run

1. Train the model and test it with sample images:

   Run the main.py script to train and evaluate the model:

   python main.py
   

   This will:

   • Train the MobileNet model on the Cats vs. Dogs dataset.
   • Save the trained model to a file mobilenet_model.h5.
   • Evaluate the model on the test set and print the test accuracy.

   The main.py script also includes a test phase that:

   • Loads the trained model.
   • Preprocesses a sample image.
   • Makes a prediction and displays the image with the predicted class label.

2. Test the model without training:

   Download your desired image and save it in the main directory of project with the name test.jpg

   # For Example
   wget -q -O test.jpg https://www.alleycat.org/wp-content/uploads/2019/03/FELV-cat.jpg
   

   Run the test.py script to test the model:

   python test.py
   

   This will:

   • Loads the trained model.
   • Preprocesses the sample image.
   • Makes a prediction and displays the image with the predicted class label.

   In this case the model uses the pre-trained weights mobilenet_model.h5 which i have provided in the repository.

Utility Functions

utils.py contains helper functions for preprocessing images:

• preprocess(image, label): Preprocesses the dataset images (resize and normalize).
• load_and_preprocess_image(image_path): Loads and preprocesses a single image for prediction.

Dataset

The dataset used for training is the Cats vs. Dogs dataset from TensorFlow Datasets.

Notes

• The model architecture and training parameters can be adjusted in model.py and train.py respectively.
• The dataset used is the Cats vs. Dogs dataset available in TensorFlow Datasets. Ensure you have a stable internet connection to download the dataset.

License

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

Acknowledgments

• The MobileNet model architecture is inspired by the original MobileNet paper.
• To implement the architecture, this post on Kaggle has been used.