LeNet-5-Visualization

Visualizing each layer output of LeNet-5 model trained on mnist data(digits)

Architecture

Official Paper

Model Summary

Trained on mnist data

Performance

loss: 0.0189 - accuracy: 0.9938 - val_loss: 0.0562 - val_accuracy: 0.9868

Summary

Visualization

Original Image

Layer-1(conv2d)

Layer-2(max_pooling2d)

Layer-3(conv2d_1)

Layer-4(max_pooling2d_1)

Try it yourself on different digits on google colab or on local system by using checkpoint.zip and LeNet-5.ipynb in the repository