Handwritten Digit Recognition with Convolutional Neural Network

A computer vision model that predicts a handwritten digit.

Convolutional Neural Network classifier for MNIST dataset featuring:

• 4 Conv2D layers with Batch Normalization
• Global Average Pooling
• 92.6% test accuracy after 1 epoch

HandwrittenImages MNIST DATABASE

🏆 Results

Metric	Score
Training Accuracy	91.6%
Test Accuracy	92.6%
Training Time	66.6s

https://www.breakthroughtech.org/programs/the-ai-program/