This example using Tensorflow Inception V3.
[Sample Images] : https://download.tensorflow.org/example_images/flower_photos.tgz
Image retrain command
<<<<<<< HEAD
$> sudo python retrain.py --image_dir flower_photos/ --output_graph ./dist/output_graph.pb
=======
$> python retrain.py --image_dir flower_photos/ --how_many_training_steps 100 --output_labels ./tensor/output_labels.txt --output_graph ./tensor/output_graph.pb --bottleneck_dir ./tensor/bottleneck --saved_model_dir ./tensor
>>>>>>> [feature] added models
#result
INFO:tensorflow:Restoring parameters from /tmp/_retrain_checkpoint
INFO:tensorflow:Froze 378 variables.
INFO:tensorflow:Converted 378 variables to const ops.
sudo tensorflowjs_converter
--input_format=tf_frozen_model
--output_node_names='final_result'
./dist/output_graph.pb
./mobilenet/web_model
$>python label_image.py --image ./test_img/sample_daisy.jpg --graph /tmp/output_graph.pb --labels /tmp/output_labels.txt
2018-12-23 08:46:37.844604: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
daisy 0.9978969
sunflowers 0.0011470014
dandelion 0.0008095624
roses 7.724412e-05
tulips 6.9350484e-05
tensorflowjs_converter \
—input_format=tf_frozen_model \
—output_node_names='final_result' \
./tensor/output_graph.pb \
./tensor/web_model
tensorflowjs_converter \
--input_format tf_frozen_model \
--output_node_names final_result \
./tensor/output_graph.pb \
./tensor/web_model
https://github.com/golbin/TensorFlow-Tutorials/tree/master/11%20-%20Inception https://www.image-net.org/ https://reiinakano.github.io/arbitrary-image-stylization-tfjs/