Handwritten Digit Recognition with Deep Learning

This project demonstrates the use of a deep learning model to recognize handwritten digits (0-9) using TensorFlow and Keras. The model is trained on a dataset of handwritten digit images and can make predictions on new images of handwritten digits.

Getting Started

Follow these instructions to get a copy of the project up and running on your local machine for testing and development purposes.

Prerequisites

To run the code in this project, you need the following dependencies installed on your system:

• Python 3.x
• TensorFlow 2.x
• NumPy
• Pillow (PIL)

Data to download

https://www.kaggle.com/datasets/scolianni/mnistasjpg/

You can install these dependencies using the following commands:

pip install tensorflow numpy pillow

Dataset

The dataset used for training this model should be organized as follows:

data/
    trainingSet/
        0/
            img_0.jpg
            img_1.jpg
            ...
        1/
            img_0.jpg
            img_1.jpg
            ...
        ...
        9/
            img_0.jpg
            img_1.jpg
            ...

Each digit class (0-9) has its own subfolder, and images are named as img_{num}.jpg.

Training the Model

1. Organize your dataset as described above.
2. Run the training script to train the model:

python trainer.py

3. The trained model will be saved as 'digit_recognition_model.h5'.

Using the Trained Model

You can use the trained model to make predictions on new handwritten digit images. Create a Python script and load the model as shown in predict_digit.py:

from tensorflow.keras.models import load_model
from PIL import Image

# Load the trained model
model = load_model('digit_recognition_model.h5')

# Preprocess an image and predict the digit
# Replace 'sample_digit.jpg' with the path to your own image
image_path = 'sample_digit.jpg'
predicted_digit = predict_digit(image_path)
print(f'Predicted Digit: {predicted_digit}')

Replace 'sample_digit.jpg' with the path to your handwritten digit image.