Image Classifier Project for AI Programming with Python Nanodegree (Udacity)
python train.py --arch densenet --hidden_units 1000,500 --epochs 10 --learning_rate 0.001 --gpu true --data_dir flowers/ --save_dir checkpoint.pth
python predict.py --checkpoint checkpoint.pth --input flowers/test/32/image_051001.jpg --top_k 5 --category_names cat_to_name.json --gpu true
To train a model, run train.py with the desired model architecture (densenet or vgg) and the path to the image folder:
python train.py --arch densenet --data_dir flowers [image folder with train, val and test sub-folders]
The command above will use default values for all other values. See below for how to customize these values.
usage: train.py [-h] [--data_dir DATA_DIR] [--save_dir SAVE_DIR] [--arch ARCH]
[--learning_rate LEARNING_RATE] [--hidden_units HIDDEN_UNITS]
[--epochs EPOCHS] [--gpu GPU]
Provide image_dir, save_dir, architecture, hyperparameters such as
learningrate, num of hidden_units, epochs and whether to use gpu or not
optional arguments:
-h, --help show this help message and exit
--data_dir DATA_DIR path to image folder
--save_dir SAVE_DIR folder where model checkpoints gets saved to
--arch ARCH choose between vgg and densenet
--learning_rate LEARNING_RATE
learning_rate for model
--hidden_units HIDDEN_UNITS
hidden_units for model
--epochs EPOCHS epochs for model
--gpu GPU whether gpu should be used for or not
To make a prediction, run predict.py with the desired checkpoint and path to the image you want to try and predict:
python predict.py --checkpoint densenet201.pth --input flowers/test/23/image_05100.jpg
The command above will use default values for all other values. See below for how to customize these values.
usage: predict.py [-h] [--input INPUT] [--checkpoint CHECKPOINT]
[--top_k TOP_K] [--category_names CATEGORY_NAMES]
[--gpu GPU]
Provide input, checkpoint, top_k, category_names and gpu
optional arguments:
-h, --help show this help message and exit
--input INPUT path to input image
--checkpoint CHECKPOINT
path to checkpoint
--top_k TOP_K number of top_k to show
--category_names CATEGORY_NAMES
path to cat names file
--gpu GPU whether gpu should be used for or not