def test_efficientnetlite0_model(self):
model = image_classifier.create(self.train_data,
model_spec.efficientnet_lite0_spec,
epochs=2,
batch_size=4,
shuffle=True)
self._test_accuracy(model)
self._test_export_to_tflite(model)
def test_efficientnetlite0_model_with_model_maker_retraining_lib(self):
model = image_classifier.create(self.train_data,
model_spec.efficientnet_lite0_spec(),
epochs=1,
batch_size=1,
shuffle=True,
use_hub_library=False)
self._test_accuracy(model)
self._test_export_to_tflite(model)
def test_resnet_50_model(self):
model = image_classifier.create(self.train_data,
mef.ModelExportFormat.TFLITE,
model_spec.resnet_50_spec,
epochs=2,
batch_size=4,
shuffle=True)
self._test_accuracy(model)
self._test_export_to_tflite(model)
def test_resnet_50_model(self):
model = image_classifier.create(self.train_data,
model_spec.resnet_50_spec(),
epochs=1,
batch_size=1,
shuffle=True)
self._test_accuracy(model)
self._test_export_to_tflite(model)
self._test_export_to_tflite_quantized(model, self.train_data)
self._test_export_to_tflite_with_metadata(model)
def test_efficientnetlite0_model(self):
model = image_classifier.create(self.train_data,
model_spec.efficientnet_lite0_spec(),
epochs=1,
batch_size=1,
shuffle=True)
self._test_accuracy(model)
self._test_export_to_tflite(model)
self._test_export_to_tflite_quantized(model, self.train_data)
self._test_export_to_tflite_with_metadata(
model, expected_json_file='efficientnet_lite0_metadata.json')
def test_mobilenetv2_model(self):
model = image_classifier.create(self.train_data,
mef.ModelExportFormat.TFLITE,
model_spec.mobilenet_v2_spec,
epochs=2,
batch_size=4,
shuffle=True)
self._test_accuracy(model)
self._test_export_to_tflite(model)
self._test_predict_top_k(model)
self._test_export_to_tflite_quantized(model, self.train_data)
def test_mobilenetv2_model(self):
model = image_classifier.create(self.train_data,
model_spec.mobilenet_v2_spec(),
epochs=1,
batch_size=1,
shuffle=True)
self._test_accuracy(model)
self._test_predict_top_k(model)
self._test_export_to_tflite(model)
self._test_export_to_tflite_quantized(model, self.train_data)
self._test_export_to_tflite_with_metadata(model)
self._test_export_to_saved_model(model)
self._test_export_labels(model)
def run(data_dir, export_dir, spec='efficientnet_lite0', **kwargs):
"""Runs demo."""
spec = model_spec.get(spec)
data = ImageClassifierDataLoader.from_folder(data_dir)
train_data, rest_data = data.split(0.8)
validation_data, test_data = rest_data.split(0.5)
model = image_classifier.create(train_data,
model_spec=spec,
validation_data=validation_data,
**kwargs)
_, acc = model.evaluate(test_data)
print('Test accuracy: %f' % acc)
model.export(export_dir)
def main(_):
logging.set_verbosity(logging.INFO)
image_path = tf.keras.utils.get_file(
'flower_photos',
'https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz',
untar=True)
data = ImageClassifierDataLoader.from_folder(image_path)
train_data, rest_data = data.split(0.8)
validation_data, test_data = rest_data.split(0.5)
model = image_classifier.create(
train_data,
model_export_format=ModelExportFormat.TFLITE,
model_spec=efficientnet_b0_spec,
validation_data=validation_data)
_, acc = model.evaluate(test_data)
print('Test accuracy: %f' % acc)
model.export(FLAGS.tflite_filename, FLAGS.label_filename)
def run(data_dir,
tflite_filename,
label_filename,
spec='efficientnet_b0',
**kwargs):
"""Runs demo."""
spec = model_spec.get(spec)
data = ImageClassifierDataLoader.from_folder(data_dir)
train_data, rest_data = data.split(0.8)
validation_data, test_data = rest_data.split(0.5)
model = image_classifier.create(
train_data,
model_export_format=ModelExportFormat.TFLITE,
model_spec=spec,
validation_data=validation_data,
**kwargs)
_, acc = model.evaluate(test_data)
print('Test accuracy: %f' % acc)
model.export(tflite_filename, label_filename)
def run(data_dir, export_dir, spec='efficientnet_lite0', **kwargs):
"""Runs demo."""
spec = model_spec.get(spec)
data = ImageClassifierDataLoader.from_folder(data_dir)
train_data, rest_data = data.split(0.8)
validation_data, test_data = rest_data.split(0.5)
model = image_classifier.create(train_data,
model_spec=spec,
validation_data=validation_data,
**kwargs)
_, acc = model.evaluate(test_data)
print('Test accuracy: %f' % acc)
# Exports to TFLite and SavedModel, with label file.
export_format = [
ExportFormat.TFLITE,
ExportFormat.SAVED_MODEL,
]
model.export(export_dir, export_format=export_format)
#from __future__ import absolute_import, division, print_function, unicode_literals
import tensorflow as tf
from tensorflow_examples.lite.model_maker.core.data_util.image_dataloader import ImageClassifierDataLoader
from tensorflow_examples.lite.model_maker.core.task import image_classifier
from tensorflow_examples.lite.model_maker.core.task.model_spec import ImageModelSpec
data = ImageClassifierDataLoader.from_folder('nsfw/')
train_data, test_data = data.split(0.8)
model = image_classifier.create(train_data)
loss, accuracy = model.evaluate(test_data)
model.export('nsfw_classifier.tflite', 'nsfw_label.txt')
def test_mobilenetv2_model_create_v1_incompatible(self):
with self.assertRaisesRegex(ValueError, 'Incompatible versions'):
_ = image_classifier.create(self.train_data,
model_spec.mobilenet_v2_spec())
def test_efficientnetlite0_model_without_training(self):
model = image_classifier.create(
self.train_data, model_spec.efficientnet_lite0_spec(), do_train=False)
self._test_accuracy(model, threshold=0.0)
self._test_export_to_tflite(model, threshold=0.0)
from tensorflow_examples.lite.model_maker.core.data_util.image_dataloader import ImageClassifierDataLoader
from tensorflow_examples.lite.model_maker.core.task import image_classifier
from tensorflow_examples.lite.model_maker.core.task import model_spec as ms
data_path = r"/home/AlgorithmicGroup/yw/workshop/antiface/data/clean_data"
# 这个path指图像数据文件夹路径，其下面按类别分为多个子文件夹
data = ImageClassifierDataLoader.from_folder(data_path)
train_data, test_data = data.split(0.92)
print("done data load.")
model = image_classifier.create(train_data,
model_spec=ms.efficientnet_lite0_spec,
shuffle=True,
validation_data=test_data,
batch_size=32,
epochs=20,
train_whole_model=False,
dropout_rate=0.2,
learning_rate=0.005,
momentum=0.9)
#指定模型为efficientnet_lite0，可以换成其他的
"""
def get_default_hparams():
return HParams(
train_epochs=5,
do_fine_tuning=False,(train_whole_model)
batch_size=32,
learning_rate=0.005,
momentum=0.9,
dropout_rate=0.2)
"""
import tensorflow as tf
from tensorflow_examples.lite.model_maker.core.data_util.image_dataloader import ImageClassifierDataLoader
from tensorflow_examples.lite.model_maker.core.task import image_classifier
from tensorflow_examples.lite.model_maker.core.task import model_spec as ms
data_path = "D:/Data/clothes_style/data/DeepFashion/classes7/data_val"
# 这个path指图像数据文件夹路径，其下面按类别分为多个子文件夹
data = ImageClassifierDataLoader.from_folder(data_path)
train_data, test_data = data.split(0.9)
print("done data load.")
model = image_classifier.create(train_data,
model_spec=ms.efficientnet_lite0_spec)
#指定模型为efficientnet_lite0，可以换成其他的
loss, accuracy = model.evaluate(test_data)
#训练过程也会打印相关信息，类似keras
model.export('image_classifier.tflite', 'image_labels.txt')
#导出tflite模型，image_labels即对应的类别