This is the 2nd project for the Deep Learning Nanodegree program at Udacity

The project built a CNN model from scratch to implement typical AlexNet architecture (including 5 Convolutional Layers and 3 fully connnected Layers, as well as some Maxpooling Layers and Dropout layers)

The project also implemented Transfer learning using Pytorch to take advantage of pre-trained VGG16 model to identify dog breeds.

In the project, dataset was preprocessed using torchvision.transforms for data augamentation. OpenCV is used for human faces detection.

The algorithms implemented in this project can be used for mobile apps or web apps.

Here is the project file ConvNet_Dog_Apps.ipynb
Note: you may use https://nbviewer.jupyter.org/ to quickly load .ipynb file