keras/resnet_v2_101_imagenet

Model Overview

Instantiates the ResNetV2 architecture. This model is supported in both KerasCV and KerasHub. KerasCV will no longer be actively developed, so please try to use KerasHub.

Reference

• Identity Mappings in Deep Residual Networks (ECCV 2016)

The difference in Resnet and ResNetV2 rests in the structure of their individual building blocks. In ResNetV2, the batch normalization and ReLU activation precede the convolution layers, as opposed to ResNetV1 where the batch normalization and ReLU activation are applied after the convolution layers.

For transfer learning use cases, make sure to read the guide to transfer learning & fine-tuning.

Links

coming soon

Installation

Keras and KerasHub can be installed with:

pip install -U -q keras-hub
pip install -U -q keras

Jax, TensorFlow, and Torch come preinstalled in Kaggle Notebooks. For instructions on installing them in another environment see the Keras Getting Started page.

Presets

The following model checkpoints are provided by the Keras team. Weights have been ported from: https://huggingface.co./timm. Full code examples for each are available below.

Preset name	Parameters	Description
resnet_v2_50_imagenet	23.56M	50-layer ResNetV2 model pre-trained on the ImageNet 1k dataset at a 224x224 resolution.
resnet_v2_101_imagenet	42.61M	101-layer ResNetV2 model pre-trained on the ImageNet 1k dataset at a 224x224 resolution.

Example Usage

    # Pretrained ResNet backbone.
    model = keras_hub.models.ResNetBackbone.from_preset("resnet_v2_101_imagenet")
    input_data = np.random.uniform(0, 1, size=(2, 224, 224, 3))
    model(input_data)

    # Randomly initialized ResNetV2 backbone with a custom config.
    model = keras_hub.models.ResNetBackbone(
        input_conv_filters=[64],
        input_conv_kernel_sizes=[7],
        stackwise_num_filters=[64, 64, 64],
        stackwise_num_blocks=[2, 2, 2],
        stackwise_num_strides=[1, 2, 2],
        block_type="basic_block",
        use_pre_activation=True,
    )
    model(input_data)
    # Use resnet for image classification task
    model = keras_hub.models.ImageClassifier.from_preset("resnet_v2_101_imagenet")

    # User timm presets directly from hugingface
    model = keras_hub.models.ImageClassifier.from_preset('hf://timm/resnetv2_101.a1h_in1k')

Example Usage with Hugging Face URI

    # Pretrained ResNet backbone.
    model = keras_hub.models.ResNetBackbone.from_preset("hf://keras/resnet_v2_101_imagenet")
    input_data = np.random.uniform(0, 1, size=(2, 224, 224, 3))
    model(input_data)

    # Randomly initialized ResNetV2 backbone with a custom config.
    model = keras_hub.models.ResNetBackbone(
        input_conv_filters=[64],
        input_conv_kernel_sizes=[7],
        stackwise_num_filters=[64, 64, 64],
        stackwise_num_blocks=[2, 2, 2],
        stackwise_num_strides=[1, 2, 2],
        block_type="basic_block",
        use_pre_activation=True,
    )
    model(input_data)
    # Use resnet for image classification task
    model = keras_hub.models.ImageClassifier.from_preset("hf://keras/resnet_v2_101_imagenet")

    # User timm presets directly from hugingface
    model = keras_hub.models.ImageClassifier.from_preset('hf://timm/resnetv2_101.a1h_in1k')