Generando imágenes con pocos datos usando S3GAN

Ver en TensorFlow.org Ejecutar en Google Colab Ver en GitHub Descargar cuaderno Ver modelos TF Hub

Este cuaderno es una demostración de las Redes Adversarias Generativas capacitadas en ImageNet con tan solo un 2,5% de datos etiquetados utilizando técnicas de aprendizaje auto-supervisado y semi-supervisado. Ambos modelos de generadores y discriminadores están disponibles en TF Hub .

Para obtener más información acerca de los modelos y el procedimiento de formación, consulte nuestra entrada de blog y el papel [1]. El código para la formación de estos modelos está disponible en GitHub .

Para comenzar, conéctese a un tiempo de ejecución y siga estos pasos:

  1. (Opcional) Seleccione un modelo en la segunda celda de código a continuación.
  2. Haga clic en tiempo de ejecución> Ejecutar toda ejecutar cada célula en orden.
    • Posteriormente, las visualizaciones interactivas deberían actualizarse automáticamente cuando modifique la configuración usando los controles deslizantes y los menús desplegables.

[1] Mario Lucic *, Michael Tschannen *, * Marvin Ritter, Xiaohua Zhai, Olivier Bachem, Sylvain Gelly, de alta fidelidad de la imagen de generación con menos etiquetas , ICML 2019.

Configuración

# @title Imports and utility functions
import os

import IPython
from IPython.display import display
import numpy as np
import PIL.Image
import pandas as pd
import six

import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()

import tensorflow_hub as hub

def imgrid(imarray, cols=8, pad=1):
  pad = int(pad)
  assert pad >= 0
  cols = int(cols)
  assert cols >= 1
  N, H, W, C = imarray.shape
  rows = int(np.ceil(N / float(cols)))
  batch_pad = rows * cols - N
  assert batch_pad >= 0
  post_pad = [batch_pad, pad, pad, 0]
  pad_arg = [[0, p] for p in post_pad]
  imarray = np.pad(imarray, pad_arg, 'constant')
  H += pad
  W += pad
  grid = (imarray
          .reshape(rows, cols, H, W, C)
          .transpose(0, 2, 1, 3, 4)
          .reshape(rows*H, cols*W, C))
  return grid[:-pad, :-pad]


def imshow(a, format='png', jpeg_fallback=True):
  a = np.asarray(a, dtype=np.uint8)
  if six.PY3:
    str_file = six.BytesIO()
  else:
    str_file = six.StringIO()
  PIL.Image.fromarray(a).save(str_file, format)
  png_data = str_file.getvalue()
  try:
    disp = display(IPython.display.Image(png_data))
  except IOError:
    if jpeg_fallback and format != 'jpeg':
      print ('Warning: image was too large to display in format "{}"; '
             'trying jpeg instead.').format(format)
      return imshow(a, format='jpeg')
    else:
      raise
  return disp


class Generator(object):

  def __init__(self, module_spec):
    self._module_spec = module_spec
    self._sess = None
    self._graph = tf.Graph()
    self._load_model()

  @property
  def z_dim(self):
    return self._z.shape[-1].value

  @property
  def conditional(self):
    return self._labels is not None

  def _load_model(self):
    with self._graph.as_default():
      self._generator = hub.Module(self._module_spec, name="gen_module",
                                   tags={"gen", "bsNone"})
      input_info = self._generator.get_input_info_dict()
      inputs = {k: tf.placeholder(v.dtype, v.get_shape().as_list(), k)
                for k, v in self._generator.get_input_info_dict().items()}
      self._samples = self._generator(inputs=inputs, as_dict=True)["generated"]
      print("Inputs:", inputs)
      print("Outputs:", self._samples)
      self._z = inputs["z"]
      self._labels = inputs.get("labels", None)

  def _init_session(self):
    if self._sess is None:
      self._sess = tf.Session(graph=self._graph)
      self._sess.run(tf.global_variables_initializer())

  def get_noise(self, num_samples, seed=None):
    if np.isscalar(seed):
      np.random.seed(seed)
      return np.random.normal(size=[num_samples, self.z_dim])
    z = np.empty(shape=(len(seed), self.z_dim), dtype=np.float32)
    for i, s in enumerate(seed):
      np.random.seed(s)
      z[i] = np.random.normal(size=[self.z_dim])
    return z

  def get_samples(self, z, labels=None):
    with self._graph.as_default():
      self._init_session()
      feed_dict = {self._z: z}
      if self.conditional:
        assert labels is not None
        assert labels.shape[0] == z.shape[0]
        feed_dict[self._labels] = labels
      samples = self._sess.run(self._samples, feed_dict=feed_dict)
      return np.uint8(np.clip(256 * samples, 0, 255))


class Discriminator(object):

  def __init__(self, module_spec):
    self._module_spec = module_spec
    self._sess = None
    self._graph = tf.Graph()
    self._load_model()

  @property
  def conditional(self):
    return "labels" in self._inputs

  @property
  def image_shape(self):
    return self._inputs["images"].shape.as_list()[1:]

  def _load_model(self):
    with self._graph.as_default():
      self._discriminator = hub.Module(self._module_spec, name="disc_module",
                                       tags={"disc", "bsNone"})
      input_info = self._discriminator.get_input_info_dict()
      self._inputs = {k: tf.placeholder(v.dtype, v.get_shape().as_list(), k)
                      for k, v in input_info.items()}
      self._outputs = self._discriminator(inputs=self._inputs, as_dict=True)
      print("Inputs:", self._inputs)
      print("Outputs:", self._outputs)

  def _init_session(self):
    if self._sess is None:
      self._sess = tf.Session(graph=self._graph)
      self._sess.run(tf.global_variables_initializer())

  def predict(self, images, labels=None):
    with self._graph.as_default():
      self._init_session()
      feed_dict = {self._inputs["images"]: images}
      if "labels" in self._inputs:
        assert labels is not None
        assert labels.shape[0] == images.shape[0]
        feed_dict[self._inputs["labels"]] = labels
      return self._sess.run(self._outputs, feed_dict=feed_dict)
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.7/site-packages/tensorflow/python/compat/v2_compat.py:111: disable_resource_variables (from tensorflow.python.ops.variable_scope) is deprecated and will be removed in a future version.
Instructions for updating:
non-resource variables are not supported in the long term

Seleccione un modelo

# @title Select a model { run: "auto" }

model_name = "S3GAN 128x128 20% labels (FID 6.9, IS 98.1)"  # @param ["S3GAN 256x256 10% labels (FID 8.8, IS 130.7)", "S3GAN 128x128 2.5% labels (FID 12.6, IS 48.7)", "S3GAN 128x128 5% labels (FID 8.4, IS 74.0)", "S3GAN 128x128 10% labels (FID 7.6, IS 90.3)", "S3GAN 128x128 20% labels (FID 6.9, IS 98.1)"]
models = {
    "S3GAN 256x256 10% labels": "https://tfhub.dev/google/compare_gan/s3gan_10_256x256/1",
    "S3GAN 128x128 2.5% labels": "https://tfhub.dev/google/compare_gan/s3gan_2_5_128x128/1",
    "S3GAN 128x128 5% labels": "https://tfhub.dev/google/compare_gan/s3gan_5_128x128/1",
    "S3GAN 128x128 10% labels": "https://tfhub.dev/google/compare_gan/s3gan_10_128x128/1",
    "S3GAN 128x128 20% labels": "https://tfhub.dev/google/compare_gan/s3gan_20_128x128/1",
}

module_spec = models[model_name.split(" (")[0]]
print("Module spec:", module_spec)

tf.reset_default_graph()
print("Loading model...")
sampler = Generator(module_spec)
print("Model loaded.")
Module spec: https://tfhub.dev/google/compare_gan/s3gan_20_128x128/1
Loading model...
INFO:tensorflow:Saver not created because there are no variables in the graph to restore
INFO:tensorflow:Saver not created because there are no variables in the graph to restore
Inputs: {'labels': <tf.Tensor 'labels:0' shape=(?,) dtype=int32>, 'z': <tf.Tensor 'z:0' shape=(?, 120) dtype=float32>}
Outputs: Tensor("gen_module_apply_default/generator_1/truediv:0", shape=(?, 128, 128, 3), dtype=float32)
Model loaded.

Muestra

png

png

Discriminado

disc = Discriminator(module_spec)

batch_size = 4
num_classes = 1000
images = np.random.random(size=[batch_size] + disc.image_shape)
labels = np.random.randint(0, num_classes, size=(batch_size))

disc.predict(images, labels=labels)
INFO:tensorflow:Saver not created because there are no variables in the graph to restore
INFO:tensorflow:Saver not created because there are no variables in the graph to restore
Inputs: {'labels': <tf.Tensor 'labels:0' shape=(?,) dtype=int32>, 'images': <tf.Tensor 'images:0' shape=(?, 128, 128, 3) dtype=float32>}
Outputs: {'prediction': <tf.Tensor 'disc_module_apply_default/discriminator/Sigmoid:0' shape=(?, 1) dtype=float32>}
{'prediction': array([[0.82321566],
        [0.89030766],
        [0.8621534 ],
        [0.88563395]], dtype=float32)}