Project code for Udacity's AI Programming with Python Nanodegree program. In this project, students first develop code for an image classifier built with PyTorch, then convert it into a command line application.
The following will train a densenet161 model executing on the GPU (Refer results_screenshot to see the obtained results. densenet161 and with 10 epochs, I am getting 95.8% accuracy.):
python train.py flower_data --arch densenet161 --learning_rate 0.001 --gpu --epochs 10
The following will train a vgg13 model executing on the GPU:
python train.py flower_data --arch vgg13 --gpu --epochs 8
The following will return the top 5 most likely classes using a pre-trained densenet161 model executing on the GPU and map categories to real names using a mapping file:
python predict.py flower_data/test/66/image_05582.jpg checkpoint_CMD_APP.pth --gpu --top_k 5 --category_names cat_to_name.json