def _buildTFGraphForName(name, featurize):
"""
Currently only supports pre-trained models from the Keras applications module.
"""
modelData = keras_apps.getKerasApplicationModel(name).getModelData(featurize)
sess = modelData["session"]
outputTensorName = modelData["outputTensorName"]
graph = tfx.strip_and_freeze_until([outputTensorName], sess.graph, sess, return_graph=True)
modelData["graph"] = graph
return modelData
def _buildTFGraphForName(name, featurize):
"""
Currently only supports pre-trained models from the Keras applications module.
"""
modelData = keras_apps.getKerasApplicationModel(name).getModelData(featurize)
sess = modelData["session"]
outputTensorName = modelData["outputTensorName"]
graph = tfx.strip_and_freeze_until([outputTensorName], sess.graph, sess, return_graph=True)
modelData["graph"] = graph
return modelData
def _buildTFGraphForName(name, featurize):
if name not in keras_apps.KERAS_APPLICATION_MODELS:
raise ValueError(
"%s is not a supported model. Supported models: %s" % name,
str(KERAS_APPLICATION_MODELS))
modelData = keras_apps.getKerasApplicationModel(name).getModelData(
featurize)
sess = modelData["session"]
outputTensorName = modelData["outputTensorName"]
graph = tfx.strip_and_freeze_until([outputTensorName],
sess.graph,
sess,
return_graph=True)
modelData["graph"] = graph
return modelData
def setUpClass(cls):
super(NamedImageTransformerBaseTestCase, cls).setUpClass()
cls.appModel = keras_apps.getKerasApplicationModel(cls.name)
imgFiles, imageArray = cls.getSampleImageList()
cls.imageArray = imageArray
cls.imgFiles = imgFiles
cls.fileOrder = {imgFiles[i].split('/')[-1]: i for i in range(len(imgFiles))}
# Predict the class probabilities for the images in our test library using keras API
# and cache for use by multiple tests.
preppedImage = cls.appModel._testPreprocess(imageArray.astype('float32'))
cls.preppedImage = preppedImage
cls.kerasPredict = cls.appModel._testKerasModel(
include_top=True).predict(preppedImage, batch_size=1)
cls.kerasFeatures = cls.appModel._testKerasModel(include_top=False).predict(preppedImage)
cls.imageDF = getSampleImageDF().limit(5)
if(cls.numPartitionsOverride):
cls.imageDf = cls.imageDF.coalesce(cls.numPartitionsOverride)
def setUpClass(cls):
super(NamedImageTransformerBaseTestCase, cls).setUpClass()
imgFiles, images = getSampleImageList()
imageArray = np.empty((len(images), 299, 299, 3), 'uint8')
for i, img in enumerate(images):
assert img is not None and img.mode == "RGB"
imageArray[i] = np.array(img.resize((299, 299)))
cls.imageArray = imageArray
# Predict the class probabilities for the images in our test library using keras API
# and cache for use by multiple tests.
cls.appModel = keras_apps.getKerasApplicationModel(cls.name)
preppedImage = cls.appModel.preprocess(imageArray.astype('float32'))
kerasPredict = cls.appModel.testKerasModel().predict(preppedImage)
cls.kerasPredict = kerasPredict
cls.imageDF = getSampleImageDF().limit(5)
def setUpClass(cls):
super(NamedImageTransformerBaseTestCase, cls).setUpClass()
cls.appModel = keras_apps.getKerasApplicationModel(cls.name)
shape = cls.appModel.inputShape()
imgFiles, images = getSampleImageList()
imageArray = np.empty((len(images), shape[0], shape[1], 3), 'uint8')
for i, img in enumerate(images):
assert img is not None and img.mode == "RGB"
imageArray[i] = np.array(img.resize(shape))
cls.imageArray = imageArray
# Predict the class probabilities for the images in our test library using keras API
# and cache for use by multiple tests.
preppedImage = cls.appModel._testPreprocess(imageArray.astype('float32'))
cls.kerasPredict = cls.appModel._testKerasModel(include_top=True).predict(preppedImage)
cls.kerasFeatures = cls.appModel._testKerasModel(include_top=False).predict(preppedImage)
cls.imageDF = getSampleImageDF().limit(5)
