October 2018
Intermediate to advanced
246 pages
6h 26m
English
With all the parameter-related details discussed, we can start the training now with the downloaded script:
python retrain.py \ --bottleneck_dir=./ \ --how_many_training_steps=4000 \ --model_dir=./ \ --output_graph=./food_graph.pb \ --output_labels=food_labels.txt \ --architecture="mobilenet_1.0_224" \ --image_dir=/Users/karthikeyan/Documents/docs/book/Chapter8/images
Based on our processor's capability as well as the number of images we have, the script might take longer for training. For me, it took more than 10 hours for 50 different food categories containing 1,500 images each. Once the script is completed, we will get the TensorFlow model in the output.
Read now
Unlock full access