CODEWITHSUNDEEP
javascripttensorflowmodelconverterstensorflow.js
Ned Bayne-Powell
Ned Bayne-Powell

Reputation: 191

How do I convert my TensorFlow saved_model.pb to Javascript?

I've been able to run the tensorflow convertor in my terminal but I'm not sure how to change the example below provided by TensorFlow to find my saved_model.pb path on the computer and save it as a Javascript file:

tensorflowjs_converter \
    --input_format=tf_saved_model \
    --output_format=tfjs_graph_model \
    --signature_name=serving_default \
    --saved_model_tags=serve \
    /mobilenet/saved_model \
    /mobilenet/web_model

Here is my input on the terminal causing the error:

tensorflowjs_converter \
    --input_format=tf_saved_model \
    --output_format=tfjs_graph_model \
    --signature_name=serving_default \
    --saved_model_tags=serve \
    /Users/name/Desktop/name\ name\ TensorFlow/saved_model  \
    /Users/name/Desktop/name\ name\ TensorFlow/web_model
usage: TensorFlow.js model converters. [-h]
                                       [--input_format {tf_frozen_model,keras_saved_model,tfjs_layers_model,tf_hub,tf_saved_model,keras}]
                                       [--output_format {tfjs_layers_model,keras_saved_model,tfjs_graph_model,keras}]
                                       [--signature_name SIGNATURE_NAME]
                                       [--saved_model_tags SAVED_MODEL_TAGS]
                                       [--quantize_float16 [QUANTIZE_FLOAT16]]
                                       [--quantize_uint8 [QUANTIZE_UINT8]]
                                       [--quantize_uint16 [QUANTIZE_UINT16]]
                                       [--quantization_bytes {1,2}]
                                       [--split_weights_by_layer] [--version]
                                       [--skip_op_check]
                                       [--strip_debug_ops STRIP_DEBUG_OPS]
                                       [--weight_shard_size_bytes WEIGHT_SHARD_SIZE_BYTES]
                                       [--output_node_names OUTPUT_NODE_NAMES]
                                       [--control_flow_v2 CONTROL_FLOW_V2]
                                       [--experiments EXPERIMENTS]
                                       [input_path] [output_path]
TensorFlow.js model converters.: error: unrecognized arguments: TensorFlow/saved_model /Users/ned/Desktop/name name TensorFlow/web_model

And when I try the Tensorflow convertor it says the original path was wrong:

Welcome to TensorFlow.js Converter.
? Please provide the path of model file or the directory that contains model fil
If you are converting TFHub module please provide the URL.  /Users/name/Desktop/name\ name\ TensorFlow/saved_model.pb 
? What is your input model format? (model format cannot be detected.)   Tensorfl
? The original path seems to be wrong, what is the directory that contains the m
odel?

Upvotes: 1

Views: 2074

Answers (2)

Andy K
Andy K

Reputation: 191

I find it easier myself to write the convertor in Python, move the script to the same folder as the directory, and then run it there. Like so:

from tensorflow import keras
import tensorflowjs as tfjs

def importModel(modelPath):
    model = keras.models.load_model(modelPath)
    tfjs.converters.save_keras_model(model, "tfjsmodel")

importModel("modelDirectory")

This way you only have to write the model's directory name in a relative pathname.

Upvotes: 2

Ned Bayne-Powell
Ned Bayne-Powell

Reputation: 191

It only allowed access to folders under home so it would just be:

/Users/ned/Tensorflow

When trying to find the path using Tensorflow Convertor.

Upvotes: 0

Related Questions