HandwrittenDigitsRecognition_DL

Deep Learning based project with Simple Feed Forward Neural Network for recognition of handwritten digits.

🖊️ Handwritten Digits Recognition (MNIST)

A simple deep learning project for recognizing handwritten digits (0–9) from the MNIST dataset. Built using TensorFlow/Keras, the model achieves high accuracy (>97%) on the test dataset.

---

📊 Dataset

• MNIST: 70,000 grayscale images of handwritten digits.

  • Training: 60,000 images
  • Testing: 10,000 images

• Each image is 28×28 pixels.

---

🧠 Model Architectures

This project explores multiple Keras Sequential Models:

1️⃣ Single Dense Layer

model = keras.Sequential([
    keras.layers.Dense(10, input_shape=(784,), activation='sigmoid')
])

2️⃣ Two-Layer Dense Network

model = keras.Sequential([
    keras.layers.Dense(100, input_shape=(784,), activation='relu'),
    keras.layers.Dense(10, activation='sigmoid')
])

3️⃣ Flatten + Dense (Best Performing)

model = keras.Sequential([
    keras.layers.Flatten(input_shape=(28, 28)),
    keras.layers.Dense(100, activation='relu'),
    keras.layers.Dense(10, activation='sigmoid')
])

---

⚙️ Installation

Clone this repository and install dependencies:

git clone https://github.com/yourusername/HandwrittenDigitsRecognition.git
cd HandwrittenDigitsRecognition
pip install -r requirements.txt

Or install manually:

pip install tensorflow matplotlib seaborn numpy

---

▶️ Usage

Run the Jupyter notebook:

jupyter notebook HandwrittenDigitsRecognition.ipynb

📈 Results

• Achieved 97%+ accuracy on the MNIST test dataset.
• Below is the confusion matrix visualization:

---

🔮 Future Work

• Use Softmax instead of Sigmoid for multi-class classification.
• Implement Convolutional Neural Networks (CNNs) for higher accuracy.
• Add Dropout/BatchNorm for better generalization.

---

🙌 Acknowledgements

• TensorFlow/Keras
• MNIST Dataset by Yann LeCun

---

📜 License

This project is licensed under the MIT License.

---