TensorFlow

TensorFlow

References:

https://www.tensorflow.org/versions/master/tutorials/layers

https://www.tensorflow.org/serving/tutorials/Serving_REST_simple

Retrain:
python retrain_ORIG.py --bottleneck_dir=/tf_files/bottlenecks --how_many_training_steps=500 --model_dir=/tf_files/inception --output_graph=/tf_files/retrained_graph.pb --output_labels=/tf_files/retrained_labels.txt --image_dir=/tf_files/train_images/ --summaries_dir=/tmp/retrain_logs

TensorBoard

tensorboard --logdir=name1:/path/to/logs/1,name2:/path/to/logs/2

Commands

!saved_model_cli show --dir {export_path} --all

Example:
saved_model_cli show --dir train_strawberrydc_iter5_labEq_combo_augmented_both_167/saved_model --all

To check versions

pip show keras
Name: Keras
Version: 2.1.2

pip show tensorflow-gpu
Name: tensorflow-gpu
Version: 1.3.0

Mask R-CNN

# How to convert COCO model (.h5) to tensorflow .pb mode
Ref: https://github.com/matterport/Mask_RCNN/issues/218

To produce h5:
python3 coco.py evaluate --dataset=$COCO_PATH --model=coco

To save model in coco.py:
evaluate_coco(model, dataset_val, coco, "bbox", limit=int(args.limit)) model.keras_model.save("mrcnn_eval.h5")

Extracting pb from h5:
python3 keras_to_tensorflow.py -input_model_file saved_model_mrcnn_eval.h5 -output_model_file model.pb -num_outputs=7