def get_model(name):
assert name == 'vgg-16'
model = keras.applications.vgg16.VGG16()
model_preprocessing = keras.applications.resnet50.preprocess_input
load_preprocess = lambda image_filepaths: model_preprocessing(
load_images(image_filepaths, image_size=224))
wrapper = KerasWrapper(model, load_preprocess)
wrapper.image_size = 224
return wrapper
def keras_model(module, model_function, image_size, identifier=None, model_kwargs=None):
module = import_module(f"keras.applications.{module}")
model_ctr, model_preprocessing = getattr(module, model_function), getattr(module, "preprocess_input")
model = model_ctr(**(model_kwargs or {}))
from model_tools.activations.keras import load_images
load_preprocess = lambda image_filepaths: model_preprocessing(load_images(image_filepaths, image_size=image_size))
wrapper = KerasWrapper(model, load_preprocess, identifier=identifier)
wrapper.image_size = image_size
return wrapper
def get_model(name):
assert name == 'resnet50'
model = TFSlimModel.init('resnet-50_v1',
net_name='resnet_v1_50',
preprocessing_type='vgg',
image_size=224,
labels_offset=0)
model_preprocessing = keras.applications.resnet50.preprocess_input
load_preprocess = lambda image_filepaths: model_preprocessing(
load_images(image_filepaths, image_size=224))
wrapper = KerasWrapper(model, load_preprocess)
wrapper.image_size = 224
return wrapper
def load_preprocess_images(image_filepaths):
images = load_images(image_filepaths)
images = [transform(image) for image in images]
images = [image.unsqueeze(0) for image in images]
images = np.concatenate(images)
return images