Views
No views yet
pip install -U -q keras-hub
pip install -U -q keras| Preset name | Parameters | Description |
|---|---|---|
| resnet_18_imagenet | 11.19M | 18-layer ResNet model pre-trained on the ImageNet 1k dataset at a 224x224 resolution |
| resnet_50_imagenet | 23.56M | 50-layer ResNet model pre-trained on the ImageNet 1k dataset at a 224x224 resolution |
| resnet_101_imagenet | 42.61M | 101-layer ResNet model pre-trained on the ImageNet 1k dataset at a 224x224 resolution |
| resnet_152_imagenet | 58.30M | 52-layer ResNet model pre-trained on the ImageNet 1k dataset at a 224x224 resolution |
1 # Pretrained ResNet backbone.
2 model = keras_hub.models.ResNetBackbone.from_preset("resnet_18_imagenet")
3 input_data = np.random.uniform(0, 1, size=(2, 224, 224, 3))
4 model(input_data)
5
6 # Randomly initialized ResNetV2 backbone with a custom config.
7 model = keras_hub.models.ResNetBackbone(
8 input_conv_filters=[64],
9 input_conv_kernel_sizes=[7],
10 stackwise_num_filters=[64, 64, 64],
11 stackwise_num_blocks=[2, 2, 2],
12 stackwise_num_strides=[1, 2, 2],
13 block_type="basic_block",
14 use_pre_activation=True,
15 )
16 model(input_data)
17 # Use resnet for image classification task
18 model = keras_hub.models.ImageClassifier.from_preset("resnet_18_imagenet")
19
20 # User timm presets directly from hugingface
21 model = keras_hub.models.ImageClassifier.from_preset('hf://timm/resnet101.a1_in1k')1 # Pretrained ResNet backbone.
2 model = keras_hub.models.ResNetBackbone.from_preset("hf://keras/resnet_18_imagenet")
3 input_data = np.random.uniform(0, 1, size=(2, 224, 224, 3))
4 model(input_data)
5
6 # Randomly initialized ResNetV2 backbone with a custom config.
7 model = keras_hub.models.ResNetBackbone(
8 input_conv_filters=[64],
9 input_conv_kernel_sizes=[7],
10 stackwise_num_filters=[64, 64, 64],
11 stackwise_num_blocks=[2, 2, 2],
12 stackwise_num_strides=[1, 2, 2],
13 block_type="basic_block",
14 use_pre_activation=True,
15 )
16 model(input_data)
17 # Use resnet for image classification task
18 model = keras_hub.models.ImageClassifier.from_preset("hf://keras/resnet_18_imagenet")
19
20 # User timm presets directly from hugingface
21 model = keras_hub.models.ImageClassifier.from_preset('hf://timm/resnet101.a1_in1k')