Python resnet.ResNet101() Examples

The following are 2 code examples of resnet.ResNet101(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module resnet , or try the search function .
Example #1
Source File: JPU.py    From TF.Keras-Commonly-used-models with Apache License 2.0 5 votes vote down vote up 
def JPU_DeepLab(img_height=1024, img_width=1024, nclasses=19):
    base_model = ResNet101(include_top=False,
                           input_shape=[img_height, img_width, 3],
                           weights=None)#'imagenet'
    endpoint_names = ['conv2_block3_out', 'conv3_block4_out',
                      'conv4_block23_out', 'conv5_block3_out']
    endpoints = [base_model.get_layer(x).output for x in endpoint_names]

    _, image_features = JPU(endpoints)

    x_a = ASPP(image_features)
    h_t, w_t = x_a.shape.as_list()[1:3]
    scale = (img_height / 4) // h_t, (img_width / 4) // w_t
    x_a = tf.keras.layers.UpSampling2D(
        size=scale, interpolation='bilinear')(x_a)

    x_b = base_model.get_layer('conv2_block3_out').output
    x_b = conv_block(x_b, num_filters=48, kernel_size=1)

    x = tf.keras.layers.Concatenate(axis=-1)([x_a, x_b])
    x = conv_block(x, num_filters=256, kernel_size=3)
    x = conv_block(x, num_filters=256, kernel_size=3)
    h_t, w_t = x.shape.as_list()[1:3]
    scale = img_height // h_t, img_width // w_t
    x = tf.keras.layers.UpSampling2D(size=scale, interpolation='bilinear')(x)

    x = tf.keras.layers.Conv2D(nclasses, (1, 1), name='output_layer')(x)
    model = tf.keras.Model(inputs=base_model.input, outputs=x, name='JPU')
    return model
Example #2
Source File: model_factory_dict.py    From mixed-precision-pytorch with Do What The F*ck You Want To Public License 4 votes vote down vote up 
def model_factory(model_name, **params):
    model_dict = {
        'densenet121': DenseNet121,
        'densenet169': DenseNet169,
        'densenet201': DenseNet201,
        'densenet161': DenseNet161,
        'densenet-cifar': densenet_cifar,
        'dual-path-net-26': DPN26,
        'dual-path-net-92': DPN92,
        'googlenet': GoogLeNet,
        'lenet': LeNet,
        'mobilenet': MobileNet,
        'mobilenetv2': MobileNetV2,
        'pnasneta': PNASNetA,
        'pnasnetb': PNASNetB,
        'preact-resnet18': PreActResNet18,
        'preact-resnet34': PreActResNet34,
        'preact-resnet50': PreActResNet50,
        'preact-resnet101': PreActResNet101,
        'preact-resnet152': PreActResNet152,
        'resnet18': ResNet18,
        'resnet34': ResNet34,
        'resnet50': ResNet50,
        'resnet101': ResNet101,
        'resnet152': ResNet152,
        'resnext29_2x64d': ResNeXt29_2x64d,
        'resnext29_4x64d': ResNeXt29_4x64d,
        'resnext29_8x64d': ResNeXt29_8x64d,
        'resnext29_32x64d': ResNeXt29_32x4d,
        'senet18': SENet18,
        'shufflenetg2': ShuffleNetG2,
        'shufflenetg3': ShuffleNetG3,
        'shufflenetv2_0.5': ShuffleNetV2,
        'shufflenetv2_1.0': ShuffleNetV2,
        'shufflenetv2_1.5': ShuffleNetV2,
        'shufflenetv2_2.0': ShuffleNetV2,
        'vgg11': VGG,
        'vgg13': VGG,
        'vgg16': VGG,
        'vgg19': VGG,
    }

    if 'vgg' in model_name:
        return model_dict[model_name](model_name)
    elif 'shufflenetv2' in model_name:
        return model_dict[model_name](float(model_name[-3:]))
    elif model_name in model_dict.keys():
        return model_dict[model_name]()
    else:
        raise AttributeError('Model doesn\'t exist')