AIPND_ImageClassifier

Udacity AI Programming with Python Nanodegree Image Classifier Project

Requirements

• python3
• matplotlib
• numpy
• pillow
• PyTorch

Training a classifier

Required Input:

• data_dir: Data Directory is expected to be in the structure defined by PyTorch ImageFolder

Optional Input:

• --save_dir: The path to write the model checkpoint file to
• --arch: The architecture to use, check PyTorch torchvision.models for available options
• --learning_rate: The learning rate to be used in the code
• --hidden_units: The input values to be applied to the hidden layers
• --epochs: The number of times to train the classifier
• --gpu: Use this flag to run with gpu

Example command

python train.py /path/to/data/dir --save_dir /path/to/output/dir --arch vgg16 --learning_rate 0.001 --hidden_units 512 256 --epochs 3 --gpu 

Making a prediction

Required Input:

• image: The path to the image to predict an output for
• checkpoint: The path to the checkpoint file to read in classifier creation

Optional Input:

• --top_k: The number of most likely classes to return
• --category_names: The path to the JSON file mapping categories to names
• --gpu: Use this flag to run with gpu

Example command

python predict.py /path/to/input/image /path/to/checkpoint/file.pth --top_k 5 --category_names /path/to/file.json --gpu