Handwritten Digit Recognizer

This project is a web application that uses a CNN model to recognize handwritten digits (0-9) drawn by users or uploaded as image files. The model predicts the digit along with the confidence level of each possible outcome, and a probablity distribution bar graph for each number. The CNN model achieved 97% accuracy and is trained and tested on MNIST data, https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz Built with React, Vite, JavaScript, Python, TensorFlow, and a backend API for model inference.

--Performance Metrics--

1, Model Performance:

• Training Accuracy: 98.6%
• Validation Accuracy: 98.12%
• Test Set Accuracy: 97.69%
• Confusion Matrix shows strong diagonal performance

2, System Performance:

• Average API Response Time: ~2.8s
• Model Size: ~6MB
• Image Processing Time: ~0.3s
• Frontend Load Time: <1s

3, Model Architecture:

Input Layer: 784 neurons (28x28 flattened)

Hidden Layers:

• Dense Layer 1: 128 neurons (ReLU)
• Dense Layer 2: 64 neurons (Sigmoid)
• Dense Layer 3: 32 neurons (Sigmoid)

Output Layer: 10 neurons (Softmax)