mnist-digit-classifier

🧠 MNIST Handwritten Digit Classifier

A simple yet effective neural network built using TensorFlow and Keras to classify handwritten digits (0–9) from the MNIST dataset.

This project serves as a beginner-friendly example of training, evaluating, and publishing a machine learning model on the Hugging Face Model Hub.

---

📌 Model Information

---

🏗️ Model Architecture

Input Layer (784)

        │

        ▼

Dense (128 neurons, ReLU)

        │

        ▼

Dense (64 neurons, ReLU)

        │

        ▼

Dense (10 neurons, Softmax)

---

📚 Dataset

The model was trained on the MNIST Handwritten Digits Dataset, containing:

• 60,000 training images
• 10,000 testing images
• Image Size: 28 × 28 pixels
• Classes: 0–9

Each image is flattened into a 784-dimensional vector before being passed into the neural network.

---

⚙️ Preprocessing

The following preprocessing steps were applied:

• Pixel values normalized to [0, 1]
• Images flattened from 28×28 → 784
• Labels encoded as integer class IDs

---

📈 Performance

The small gap between training and test accuracy indicates that the model generalizes reasonably well on unseen data.

---

🚀 Loading the Model

import tensorflow as tf

model = tf.keras.models.load_model("mnist_classifier.keras")

---

🔍 Making Predictions

import numpy as np
import tensorflow as tf

model = tf.keras.models.load_model("mnist_classifier.keras")

# image should be normalized and reshaped
image = image.reshape(1, 784)

prediction = model.predict(image)

predicted_digit = np.argmax(prediction)

print(predicted_digit)

---

📦 Requirements

• Python 3.10+
• TensorFlow
• NumPy

Install dependencies:

pip install tensorflow numpy

---

📁 Repository Contents

.
├── mnist_classifier.keras
├── README.md

---

🎯 Intended Use

This model is intended for:

• Learning TensorFlow and Keras
• Educational purposes
• MNIST handwritten digit classification
• Demonstrating model deployment on Hugging Face

---

⚠️ Limitations

• Designed only for MNIST-style handwritten digits.
• Expects grayscale images of handwritten digits.
• Images should be preprocessed in the same way as the training data.

---

📄 License

This project is released under the MIT License, allowing free use, modification, and distribution.

---

👨‍💻 Author

Madhav Maheshwari

If you find this project useful, consider giving it a ⭐ on Hugging Face.