add application_efficientnet_b{0:7}()

Author: t-kalinowski

NAMESPACE

@@ -93,6 +93,14 @@ export(application_densenet)
 export(application_densenet121)
 export(application_densenet169)
 export(application_densenet201)
+export(application_efficientnet_b0)
+export(application_efficientnet_b1)
+export(application_efficientnet_b2)
+export(application_efficientnet_b3)
+export(application_efficientnet_b4)
+export(application_efficientnet_b5)
+export(application_efficientnet_b6)
+export(application_efficientnet_b7)
 export(application_inception_resnet_v2)
 export(application_inception_v3)
 export(application_mobilenet)

NEWS.md

@@ -26,6 +26,8 @@
 
 - New vignette: "Transfer learning and fine-tuning".
 
+- New `application_efficientnet_b{1,2,3,4,5,6,7}()`.
+
 - New function `%<-active%`, a ergonomic wrapper around `makeActiveBinding()`
   for constructing Python `@property` decorated methods in `%py_class%`.
 

R/applications.R

@@ -692,6 +692,178 @@ application_nasnetmobile <- function(input_shape = NULL, include_top = TRUE, wei
 
 }
 
+#' Instantiates the EfficientNetB0 architecture
+#'
+#' @details
+#' Reference:
+#' - [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](
+#'     https://arxiv.org/abs/1905.11946) (ICML 2019)
+#'
+#' This function returns a Keras image classification model,
+#' optionally loaded with weights pre-trained on ImageNet.
+#'
+#' For image classification use cases, see
+#' [this page for detailed examples](
+#'   https://keras.io/api/applications/#usage-examples-for-image-classification-models).
+#'
+#' For transfer learning use cases, make sure to read the
+#' [guide to transfer learning & fine-tuning](
+#'   https://keras.io/guides/transfer_learning/).
+#'
+#' EfficientNet models expect their inputs to be float tensors of pixels with values in the [0-255] range.
+#'
+#' @note
+#' Each Keras Application typically expects a specific kind of input preprocessing.
+#' For EfficientNet, input preprocessing is included as part of the model
+#' (as a `Rescaling` layer), and thus a calling a preprocessing function is not necessary.
+#'
+#'
+#' @param include_top Whether to include the fully-connected
+#' layer at the top of the network. Defaults to TRUE.
+#'
+#' @param weights One of `NULL` (random initialization),
+#' `'imagenet'` (pre-training on ImageNet),
+#' or the path to the weights file to be loaded. Defaults to `'imagenet'`.
+#'
+#' @param input_tensor Optional Keras tensor
+#' (i.e. output of `layer_input()`)
+#' to use as image input for the model.
+#'
+#' @param input_shape Optional shape list, only to be specified
+#' if `include_top` is FALSE.
+#' It should have exactly 3 inputs channels.
+#'
+#' @param pooling Optional pooling mode for feature extraction
+#' when `include_top` is `FALSE`. Defaults to `NULL`.
+#' - `NULL` means that the output of the model will be
+#'     the 4D tensor output of the
+#'     last convolutional layer.
+#' - `'avg'` means that global average pooling
+#'     will be applied to the output of the
+#'     last convolutional layer, and thus
+#'     the output of the model will be a 2D tensor.
+#' - `'max'` means that global max pooling will
+#'     be applied.
+#'
+#' @param classes Optional number of classes to classify images into, only to be
+#'   specified if `include_top` is TRUE, and if no `weights` argument is
+#'   specified. Defaults to 1000 (number of ImageNet classes).
+#'
+#' @param classifier_activation A string or callable. The activation function to
+#'   use on the "top" layer. Ignored unless `include_top = TRUE`. Set
+#'   `classifier_activation = NULL` to return the logits of the "top" layer.
+#'   Defaults to `'softmax'`. When loading pretrained weights,
+#'   `classifier_activation` can only be `NULL` or `"softmax"`.
+#'
+#' @param ... for forward compatibility.
+#'
+#' @seealso
+#'   +  <https://www.tensorflow.org/api_docs/python/tf/keras/applications/efficientnet/EfficientNetB0>
+#'   +  <https://keras.io/api/applications/>
+#' @export
+application_efficientnet_b0 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB0, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b1 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB1, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b2 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB2, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b3 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB3, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b4 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB4, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b5 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB5, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b6 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB6, args)
+}
+
+#' @export
+#' @rdname application_efficientnet_b0
+application_efficientnet_b7 <-
+function(include_top = TRUE, weights = "imagenet",
+         input_tensor = NULL, input_shape = NULL,
+         pooling = NULL, classes = 1000L,
+         classifier_activation = "softmax",
+         ...)
+{
+    args <- capture_args(match.call(), list(classes = as.integer, input_shape = normalize_shape))
+    do.call(keras$applications$EfficientNetB7, args)
+}
+
+
 #' @rdname application_nasnet
 #' @export
 nasnet_preprocess_input <- function(x) {

man/application_efficientnet_b0.Rd

@@ -32,6 +32,16 @@ test_succeeds("keras pre-built models can be instantiated", {
   application_vgg16()
   application_vgg19()
   application_inception_v3()
+
+  application_efficientnet_b0()
+  application_efficientnet_b1()
+  application_efficientnet_b2()
+  application_efficientnet_b3()
+  application_efficientnet_b4()
+  application_efficientnet_b5()
+  application_efficientnet_b6()
+  application_efficientnet_b7()
+
 })
 
 test_succeeds("can use any input_shape", {