To try the default classifier model on unsplit directory of subdir-labelled images:
python main.py --image_dir "<sub-dir-labeled-image-dir>" \
--logdir "<output_dir>"
To try the default classifier model on directory of subdir-labelled images already split into train/test or train/test/val directories:
python main.py --split_image_dir "<sub-dir-labeled-split-image-dir>" \
--logdir "<output_dir>"
To try many models
for m in $(python available_tf_hub_models.py); do
python main.py --image_dir "<sub-dir-labeled-image-dir>" \
--logdir "<output_dir>" \
--run_name "$m" \
--model "$m"
done