The MobileNet model was proposed in MobileNetV2: Inverted Residuals and Linear Bottlenecks by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
The abstract from the paper is the following:
In this paper we describe a new mobile architecture, MobileNetV2, that improves the state of the art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes. We also describe efficient ways of applying these mobile models to object detection in a novel framework we call SSDLite. Additionally, we demonstrate how to build mobile semantic segmentation models through a reduced form of DeepLabv3 which we call Mobile DeepLabv3.
The MobileNetV2 architecture is based on an inverted residual structure where the input and output of the residual block are thin bottleneck layers opposite to traditional residual models which use expanded representations in the input an MobileNetV2 uses lightweight depthwise convolutions to filter features in the intermediate expansion layer. Additionally, we find that it is important to remove non-linearities in the narrow layers in order to maintain representational power. We demonstrate that this improves performance and provide an intuition that led to this design. Finally, our approach allows decoupling of the input/output domains from the expressiveness of the transformation, which provides a convenient framework for further analysis. We measure our performance on Imagenet classification, COCO object detection, VOC image segmentation. We evaluate the trade-offs between accuracy, and number of operations measured by multiply-adds (MAdd), as well as the number of parameters.
This model was contributed by matthijs. The original code and weights can be found here for the main model and here for DeepLabV3+.
The checkpoints are named mobilenet_v2_depth_size, for example mobilenet_v2_1.0_224, where 1.0 is the depth multiplier (sometimes also referred to as “alpha” or the width multiplier) and 224 is the resolution of the input images the model was trained on.
Even though the checkpoint is trained on images of specific size, the model will work on images of any size. The smallest supported image size is 32x32.
One can use MobileNetV2ImageProcessor to prepare images for the model.
The available image classification checkpoints are pre-trained on ImageNet-1k (also referred to as ILSVRC 2012, a collection of 1.3 million images and 1,000 classes). However, the model predicts 1001 classes: the 1000 classes from ImageNet plus an extra “background” class (index 0).
The segmentation model uses a DeepLabV3+ head. The available semantic segmentation checkpoints are pre-trained on PASCAL VOC.
The original TensorFlow checkpoints use different padding rules than PyTorch, requiring the model to determine the padding amount at inference time, since this depends on the input image size. To use native PyTorch padding behavior, create a MobileNetV2Config with tf_padding = False.
Unsupported features:
The MobileNetV2Model outputs a globally pooled version of the last hidden state. In the original model it is possible to use an average pooling layer with a fixed 7x7 window and stride 1 instead of global pooling. For inputs that are larger than the recommended image size, this gives a pooled output that is larger than 1x1. The Hugging Face implementation does not support this.
The original TensorFlow checkpoints include quantized models. We do not support these models as they include additional “FakeQuantization” operations to unquantize the weights.
It’s common to extract the output from the expansion layers at indices 10 and 13, as well as the output from the final 1x1 convolution layer, for downstream purposes. Using output_hidden_states=True returns the output from all intermediate layers. There is currently no way to limit this to specific layers.
The DeepLabV3+ segmentation head does not use the final convolution layer from the backbone, but this layer gets computed anyway. There is currently no way to tell MobileNetV2Model up to which layer it should run.
A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with MobileNetV2.
Semantic segmentation
If you’re interested in submitting a resource to be included here, please feel free to open a Pull Request and we’ll review it! The resource should ideally demonstrate something new instead of duplicating an existing resource.
( num_channels = 3 image_size = 224 depth_multiplier = 1.0 depth_divisible_by = 8 min_depth = 8 expand_ratio = 6.0 output_stride = 32 first_layer_is_expansion = True finegrained_output = True hidden_act = 'relu6' tf_padding = True classifier_dropout_prob = 0.8 initializer_range = 0.02 layer_norm_eps = 0.001 semantic_loss_ignore_index = 255 **kwargs )
Parameters
int, optional, defaults to 3) —
The number of input channels. int, optional, defaults to 224) —
The size (resolution) of each image. float, optional, defaults to 1.0) —
Shrinks or expands the number of channels in each layer. Default is 1.0, which starts the network with 32
channels. This is sometimes also called “alpha” or “width multiplier”. int, optional, defaults to 8) —
The number of channels in each layer will always be a multiple of this number. int, optional, defaults to 8) —
All layers will have at least this many channels. float, optional, defaults to 6.0) —
The number of output channels of the first layer in each block is input channels times expansion ratio. int, optional, defaults to 32) —
The ratio between the spatial resolution of the input and output feature maps. By default the model reduces
the input dimensions by a factor of 32. If output_stride is 8 or 16, the model uses dilated convolutions
on the depthwise layers instead of regular convolutions, so that the feature maps never become more than 8x
or 16x smaller than the input image. bool, optional, defaults to True) —
True if the very first convolution layer is also the expansion layer for the first expansion block. bool, optional, defaults to True) —
If true, the number of output channels in the final convolution layer will stay large (1280) even if
depth_multiplier is less than 1. str or function, optional, defaults to "relu6") —
The non-linear activation function (function or string) in the Transformer encoder and convolution layers. bool, optional, defaults to True) —
Whether to use TensorFlow padding rules on the convolution layers. float, optional, defaults to 0.8) —
The dropout ratio for attached classifiers. float, optional, defaults to 0.02) —
The standard deviation of the truncated_normal_initializer for initializing all weight matrices. float, optional, defaults to 0.001) —
The epsilon used by the layer normalization layers. int, optional, defaults to 255) —
The index that is ignored by the loss function of the semantic segmentation model. This is the configuration class to store the configuration of a MobileNetV2Model. It is used to instantiate a MobileNetV2 model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the MobileNetV2 google/mobilenet_v2_1.0_224 architecture.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
Example:
>>> from transformers import MobileNetV2Config, MobileNetV2Model
>>> # Initializing a "mobilenet_v2_1.0_224" style configuration
>>> configuration = MobileNetV2Config()
>>> # Initializing a model from the "mobilenet_v2_1.0_224" style configuration
>>> model = MobileNetV2Model(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config( images: Union do_resize: Optional = None size: Dict = None resample: Resampling = None do_center_crop: bool = None crop_size: Dict = None do_rescale: Optional = None rescale_factor: Optional = None do_normalize: Optional = None image_mean: Union = None image_std: Union = None return_tensors: Union = None data_format: Union = <ChannelDimension.FIRST: 'channels_first'> input_data_format: Union = None **kwargs )
Parameters
ImageInput) —
Image to preprocess. Expects a single or batch of images with pixel values ranging from 0 to 255. If
passing in images with pixel values between 0 and 1, set do_rescale=False. bool, optional, defaults to self.do_resize) —
Whether to resize the image. Dict[str, int], optional, defaults to self.size) —
Size of the image after resizing. Shortest edge of the image is resized to size[“shortest_edge”], with
the longest edge resized to keep the input aspect ratio. PILImageResampling filter, optional, defaults to self.resample) —
PILImageResampling filter to use if resizing the image e.g. PILImageResampling.BILINEAR. Only has
an effect if do_resize is set to True. bool, optional, defaults to self.do_center_crop) —
Whether to center crop the image. Dict[str, int], optional, defaults to self.crop_size) —
Size of the center crop. Only has an effect if do_center_crop is set to True. bool, optional, defaults to self.do_rescale) —
Whether to rescale the image values between [0 - 1]. float, optional, defaults to self.rescale_factor) —
Rescale factor to rescale the image by if do_rescale is set to True. bool, optional, defaults to self.do_normalize) —
Whether to normalize the image. float or List[float], optional, defaults to self.image_mean) —
Image mean to use if do_normalize is set to True. float or List[float], optional, defaults to self.image_std) —
Image standard deviation to use if do_normalize is set to True. str or TensorType, optional) —
The type of tensors to return. Can be one of:np.ndarray.TensorType.TENSORFLOW or 'tf': Return a batch of type tf.Tensor.TensorType.PYTORCH or 'pt': Return a batch of type torch.Tensor.TensorType.NUMPY or 'np': Return a batch of type np.ndarray.TensorType.JAX or 'jax': Return a batch of type jax.numpy.ndarray.ChannelDimension or str, optional, defaults to ChannelDimension.FIRST) —
The channel dimension format for the output image. Can be one of:"channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format."channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format.ChannelDimension or str, optional) —
The channel dimension format for the input image. If unset, the channel dimension format is inferred
from the input image. Can be one of:"channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format."channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format."none" or ChannelDimension.NONE: image in (height, width) format.Preprocess an image or batch of images.
( outputs target_sizes: List = None ) → semantic_segmentation
Parameters
List[Tuple] of length batch_size, optional) —
List of tuples corresponding to the requested final size (height, width) of each prediction. If unset,
predictions will not be resized. Returns
semantic_segmentation
List[torch.Tensor] of length batch_size, where each item is a semantic
segmentation map of shape (height, width) corresponding to the target_sizes entry (if target_sizes is
specified). Each entry of each torch.Tensor correspond to a semantic class id.
Converts the output of MobileNetV2ForSemanticSegmentation into semantic segmentation maps. Only supports PyTorch.
( do_resize: bool = True size: Optional = None resample: Resampling = <Resampling.BILINEAR: 2> do_center_crop: bool = True crop_size: Dict = None do_rescale: bool = True rescale_factor: Union = 0.00392156862745098 do_normalize: bool = True image_mean: Union = None image_std: Union = None **kwargs )
Parameters
bool, optional, defaults to True) —
Whether to resize the image’s (height, width) dimensions to the specified size. Can be overridden by
do_resize in the preprocess method. Dict[str, int] optional, defaults to {"shortest_edge" -- 256}):
Size of the image after resizing. The shortest edge of the image is resized to size[“shortest_edge”], with
the longest edge resized to keep the input aspect ratio. Can be overridden by size in the preprocess
method. PILImageResampling, optional, defaults to PILImageResampling.BILINEAR) —
Resampling filter to use if resizing the image. Can be overridden by the resample parameter in the
preprocess method. bool, optional, defaults to True) —
Whether to center crop the image. If the input size is smaller than crop_size along any edge, the image
is padded with 0’s and then center cropped. Can be overridden by the do_center_crop parameter in the
preprocess method. Dict[str, int], optional, defaults to {"height" -- 224, "width": 224}):
Desired output size when applying center-cropping. Only has an effect if do_center_crop is set to True.
Can be overridden by the crop_size parameter in the preprocess method. bool, optional, defaults to True) —
Whether to rescale the image by the specified scale rescale_factor. Can be overridden by the do_rescale
parameter in the preprocess method. int or float, optional, defaults to 1/255) —
Scale factor to use if rescaling the image. Can be overridden by the rescale_factor parameter in the
preprocess method.
do_normalize —
Whether to normalize the image. Can be overridden by the do_normalize parameter in the preprocess
method. float or List[float], optional, defaults to IMAGENET_STANDARD_MEAN) —
Mean to use if normalizing the image. This is a float or list of floats the length of the number of
channels in the image. Can be overridden by the image_mean parameter in the preprocess method. float or List[float], optional, defaults to IMAGENET_STANDARD_STD) —
Standard deviation to use if normalizing the image. This is a float or list of floats the length of the
number of channels in the image. Can be overridden by the image_std parameter in the preprocess method. Constructs a MobileNetV2 image processor.
( images: Union do_resize: Optional = None size: Dict = None resample: Resampling = None do_center_crop: bool = None crop_size: Dict = None do_rescale: Optional = None rescale_factor: Optional = None do_normalize: Optional = None image_mean: Union = None image_std: Union = None return_tensors: Union = None data_format: Union = <ChannelDimension.FIRST: 'channels_first'> input_data_format: Union = None **kwargs )
Parameters
ImageInput) —
Image to preprocess. Expects a single or batch of images with pixel values ranging from 0 to 255. If
passing in images with pixel values between 0 and 1, set do_rescale=False. bool, optional, defaults to self.do_resize) —
Whether to resize the image. Dict[str, int], optional, defaults to self.size) —
Size of the image after resizing. Shortest edge of the image is resized to size[“shortest_edge”], with
the longest edge resized to keep the input aspect ratio. PILImageResampling filter, optional, defaults to self.resample) —
PILImageResampling filter to use if resizing the image e.g. PILImageResampling.BILINEAR. Only has
an effect if do_resize is set to True. bool, optional, defaults to self.do_center_crop) —
Whether to center crop the image. Dict[str, int], optional, defaults to self.crop_size) —
Size of the center crop. Only has an effect if do_center_crop is set to True. bool, optional, defaults to self.do_rescale) —
Whether to rescale the image values between [0 - 1]. float, optional, defaults to self.rescale_factor) —
Rescale factor to rescale the image by if do_rescale is set to True. bool, optional, defaults to self.do_normalize) —
Whether to normalize the image. float or List[float], optional, defaults to self.image_mean) —
Image mean to use if do_normalize is set to True. float or List[float], optional, defaults to self.image_std) —
Image standard deviation to use if do_normalize is set to True. str or TensorType, optional) —
The type of tensors to return. Can be one of:np.ndarray.TensorType.TENSORFLOW or 'tf': Return a batch of type tf.Tensor.TensorType.PYTORCH or 'pt': Return a batch of type torch.Tensor.TensorType.NUMPY or 'np': Return a batch of type np.ndarray.TensorType.JAX or 'jax': Return a batch of type jax.numpy.ndarray.ChannelDimension or str, optional, defaults to ChannelDimension.FIRST) —
The channel dimension format for the output image. Can be one of:"channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format."channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format.ChannelDimension or str, optional) —
The channel dimension format for the input image. If unset, the channel dimension format is inferred
from the input image. Can be one of:"channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format."channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format."none" or ChannelDimension.NONE: image in (height, width) format.Preprocess an image or batch of images.
( outputs target_sizes: List = None ) → semantic_segmentation
Parameters
List[Tuple] of length batch_size, optional) —
List of tuples corresponding to the requested final size (height, width) of each prediction. If unset,
predictions will not be resized. Returns
semantic_segmentation
List[torch.Tensor] of length batch_size, where each item is a semantic
segmentation map of shape (height, width) corresponding to the target_sizes entry (if target_sizes is
specified). Each entry of each torch.Tensor correspond to a semantic class id.
Converts the output of MobileNetV2ForSemanticSegmentation into semantic segmentation maps. Only supports PyTorch.
( config: MobileNetV2Config add_pooling_layer: bool = True )
Parameters
The bare MobileNetV2 model outputting raw hidden-states without any specific head on top. This model is a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
( pixel_values: Optional = None output_hidden_states: Optional = None return_dict: Optional = None ) → transformers.modeling_outputs.BaseModelOutputWithPoolingAndNoAttention or tuple(torch.FloatTensor)
Parameters
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using AutoImageProcessor. See
MobileNetV2ImageProcessor.call() for details. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. Returns
transformers.modeling_outputs.BaseModelOutputWithPoolingAndNoAttention or tuple(torch.FloatTensor)
A transformers.modeling_outputs.BaseModelOutputWithPoolingAndNoAttention or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (MobileNetV2Config) and inputs.
last_hidden_state (torch.FloatTensor of shape (batch_size, num_channels, height, width)) — Sequence of hidden-states at the output of the last layer of the model.
pooler_output (torch.FloatTensor of shape (batch_size, hidden_size)) — Last layer hidden-state after a pooling operation on the spatial dimensions.
hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
one for the output of each layer) of shape (batch_size, num_channels, height, width).
Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
The MobileNetV2Model forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Example:
>>> from transformers import AutoImageProcessor, MobileNetV2Model
>>> import torch
>>> from datasets import load_dataset
>>> dataset = load_dataset("huggingface/cats-image")
>>> image = dataset["test"]["image"][0]
>>> image_processor = AutoImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224")
>>> model = MobileNetV2Model.from_pretrained("google/mobilenet_v2_1.0_224")
>>> inputs = image_processor(image, return_tensors="pt")
>>> with torch.no_grad():
... outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
>>> list(last_hidden_states.shape)
[1, 1280, 7, 7]( config: MobileNetV2Config )
Parameters
MobileNetV2 model with an image classification head on top (a linear layer on top of the pooled features), e.g. for ImageNet.
This model is a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
( pixel_values: Optional = None output_hidden_states: Optional = None labels: Optional = None return_dict: Optional = None ) → transformers.modeling_outputs.ImageClassifierOutputWithNoAttention or tuple(torch.FloatTensor)
Parameters
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using AutoImageProcessor. See
MobileNetV2ImageProcessor.call() for details. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. torch.LongTensor of shape (batch_size,), optional) —
Labels for computing the image classification/regression loss. Indices should be in [0, ..., config.num_labels - 1]. If config.num_labels == 1 a regression loss is computed (Mean-Square loss). If
config.num_labels > 1 a classification loss is computed (Cross-Entropy). Returns
transformers.modeling_outputs.ImageClassifierOutputWithNoAttention or tuple(torch.FloatTensor)
A transformers.modeling_outputs.ImageClassifierOutputWithNoAttention or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (MobileNetV2Config) and inputs.
torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Classification (or regression if config.num_labels==1) loss.torch.FloatTensor of shape (batch_size, config.num_labels)) — Classification (or regression if config.num_labels==1) scores (before SoftMax).tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
one for the output of each stage) of shape (batch_size, num_channels, height, width). Hidden-states (also
called feature maps) of the model at the output of each stage.The MobileNetV2ForImageClassification forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Example:
>>> from transformers import AutoImageProcessor, MobileNetV2ForImageClassification
>>> import torch
>>> from datasets import load_dataset
>>> dataset = load_dataset("huggingface/cats-image")
>>> image = dataset["test"]["image"][0]
>>> image_processor = AutoImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224")
>>> model = MobileNetV2ForImageClassification.from_pretrained("google/mobilenet_v2_1.0_224")
>>> inputs = image_processor(image, return_tensors="pt")
>>> with torch.no_grad():
... logits = model(**inputs).logits
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_label = logits.argmax(-1).item()
>>> print(model.config.id2label[predicted_label])
tabby, tabby cat( config: MobileNetV2Config )
Parameters
MobileNetV2 model with a semantic segmentation head on top, e.g. for Pascal VOC.
This model is a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
( pixel_values: Optional = None labels: Optional = None output_hidden_states: Optional = None return_dict: Optional = None ) → transformers.modeling_outputs.SemanticSegmenterOutput or tuple(torch.FloatTensor)
Parameters
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using AutoImageProcessor. See
MobileNetV2ImageProcessor.call() for details. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. torch.LongTensor of shape (batch_size, height, width), optional) —
Ground truth semantic segmentation maps for computing the loss. Indices should be in [0, ..., config.num_labels - 1]. If config.num_labels > 1, a classification loss is computed (Cross-Entropy). Returns
transformers.modeling_outputs.SemanticSegmenterOutput or tuple(torch.FloatTensor)
A transformers.modeling_outputs.SemanticSegmenterOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (MobileNetV2Config) and inputs.
loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Classification (or regression if config.num_labels==1) loss.
logits (torch.FloatTensor of shape (batch_size, config.num_labels, logits_height, logits_width)) — Classification scores for each pixel.
The logits returned do not necessarily have the same size as the pixel_values passed as inputs. This is
to avoid doing two interpolations and lose some quality when a user needs to resize the logits to the
original image size as post-processing. You should always check your logits shape and resize as needed.
hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
one for the output of each layer) of shape (batch_size, patch_size, hidden_size).
Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, patch_size, sequence_length).
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
The MobileNetV2ForSemanticSegmentation forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Examples:
>>> from transformers import AutoImageProcessor, MobileNetV2ForSemanticSegmentation
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> image_processor = AutoImageProcessor.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513")
>>> model = MobileNetV2ForSemanticSegmentation.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513")
>>> inputs = image_processor(images=image, return_tensors="pt")
>>> with torch.no_grad():
... outputs = model(**inputs)
>>> # logits are of shape (batch_size, num_labels, height, width)
>>> logits = outputs.logits