brails.processors.nfloors_detector.lib.efficientnet.utils module

This file contains helper functions for building the model and for loading model parameters. These helper functions are built to mirror those in the official TensorFlow implementation.

class brails.processors.nfloors_detector.lib.efficientnet.utils.BlockArgs(kernel_size, num_repeat, input_filters, output_filters, expand_ratio, id_skip, stride, se_ratio)

Bases: tuple

expand_ratio: Alias for field number 4

id_skip: Alias for field number 5

input_filters: Alias for field number 2

kernel_size: Alias for field number 0

num_repeat: Alias for field number 1

output_filters: Alias for field number 3

se_ratio: Alias for field number 7

stride: Alias for field number 6

class brails.processors.nfloors_detector.lib.efficientnet.utils.BlockDecoder

Bases: object

Block Decoder for readability, straight from the official TensorFlow repository

static decode(string_list)

Decodes a list of string notations to specify blocks inside the network.

Parameters:: string_list – a list of strings, each string is a notation of block
Returns:: a list of BlockArgs namedtuples of block args

static encode(blocks_args)

Encodes a list of BlockArgs to a list of strings.

Parameters:: blocks_args – a list of BlockArgs namedtuples of block args
Returns:: a list of strings, each string is a notation of block

class brails.processors.nfloors_detector.lib.efficientnet.utils.Conv2dDynamicSamePadding(in_channels, out_channels, kernel_size, stride=1, dilation=1, groups=1, bias=True)

Bases: Conv2d

2D Convolutions like TensorFlow, for a dynamic image size

forward(x)

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

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 registered hooks while the latter silently ignores them.

class brails.processors.nfloors_detector.lib.efficientnet.utils.GlobalParams(batch_norm_momentum, batch_norm_epsilon, dropout_rate, num_classes, width_coefficient, depth_coefficient, depth_divisor, min_depth, drop_connect_rate, image_size)

Bases: tuple

batch_norm_epsilon: Alias for field number 1

batch_norm_momentum: Alias for field number 0

depth_coefficient: Alias for field number 5

depth_divisor: Alias for field number 6

drop_connect_rate: Alias for field number 8

dropout_rate: Alias for field number 2

image_size: Alias for field number 9

min_depth: Alias for field number 7

num_classes: Alias for field number 3

width_coefficient: Alias for field number 4

class brails.processors.nfloors_detector.lib.efficientnet.utils.Identity

Bases: Module

forward(input)

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

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 registered hooks while the latter silently ignores them.

class brails.processors.nfloors_detector.lib.efficientnet.utils.MemoryEfficientSwish(*args, **kwargs)

Bases: Module

forward(x)

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

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 registered hooks while the latter silently ignores them.

class brails.processors.nfloors_detector.lib.efficientnet.utils.Swish(*args, **kwargs)

Bases: Module

forward(x)

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

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 registered hooks while the latter silently ignores them.

class brails.processors.nfloors_detector.lib.efficientnet.utils.SwishImplementation(*args, **kwargs)

Bases: Function

static backward(ctx, grad_output)

Define a formula for differentiating the operation with backward mode automatic differentiation.

This function is to be overridden by all subclasses. (Defining this function is equivalent to defining the vjp function.)

It must accept a context ctx as the first argument, followed by as many outputs as the forward() returned (None will be passed in for non tensor outputs of the forward function), and it should return as many tensors, as there were inputs to forward(). Each argument is the gradient w.r.t the given output, and each returned value should be the gradient w.r.t. the corresponding input. If an input is not a Tensor or is a Tensor not requiring grads, you can just pass None as a gradient for that input.

The context can be used to retrieve tensors saved during the forward pass. It also has an attribute ctx.needs_input_grad as a tuple of booleans representing whether each input needs gradient. E.g., backward() will have ctx.needs_input_grad[0] = True if the first input to forward() needs gradient computed w.r.t. the output.

static forward(ctx, i)

Define the forward of the custom autograd Function.

This function is to be overridden by all subclasses. There are two ways to define forward:

Usage 1 (Combined forward and ctx):

@staticmethod
def forward(ctx: Any, *args: Any, **kwargs: Any) -> Any:
    pass

It must accept a context ctx as the first argument, followed by any number of arguments (tensors or other types).
See combining-forward-context for more details

Usage 2 (Separate forward and ctx):

@staticmethod
def forward(*args: Any, **kwargs: Any) -> Any:
    pass

@staticmethod
def setup_context(ctx: Any, inputs: Tuple[Any, ...], output: Any) -> None:
    pass

The forward no longer accepts a ctx argument.
Instead, you must also override the torch.autograd.Function.setup_context() staticmethod to handle setting up the ctx object. output is the output of the forward, inputs are a Tuple of inputs to the forward.
See extending-autograd for more details

The context can be used to store arbitrary data that can be then retrieved during the backward pass. Tensors should not be stored directly on ctx (though this is not currently enforced for backward compatibility). Instead, tensors should be saved either with ctx.save_for_backward() if they are intended to be used in backward (equivalently, vjp) or ctx.save_for_forward() if they are intended to be used for in jvp.

brails.processors.nfloors_detector.lib.efficientnet.utils.drop_connect(inputs, p, training): Drop connect.

brails.processors.nfloors_detector.lib.efficientnet.utils.efficientnet(width_coefficient=None, depth_coefficient=None, dropout_rate=0.2, drop_connect_rate=0.2, image_size=None, num_classes=1000): Creates a efficientnet model.

brails.processors.nfloors_detector.lib.efficientnet.utils.efficientnet_params(model_name): Map EfficientNet model name to parameter coefficients.

brails.processors.nfloors_detector.lib.efficientnet.utils.get_model_params(model_name, override_params): Get the block args and global params for a given model

brails.processors.nfloors_detector.lib.efficientnet.utils.get_same_padding_conv2d(image_size=None): Chooses static padding if you have specified an image size, and dynamic padding otherwise. Static padding is necessary for ONNX exporting of models.

brails.processors.nfloors_detector.lib.efficientnet.utils.load_pretrained_weights(model, model_name, load_fc=True, advprop=False): Loads pretrained weights, and downloads if loading for the first time.

brails.processors.nfloors_detector.lib.efficientnet.utils.round_filters(filters, global_params): Calculate and round number of filters based on depth multiplier.

brails.processors.nfloors_detector.lib.efficientnet.utils.round_repeats(repeats, global_params): Round number of filters based on depth multiplier.