首页分享 tensorflow2的U

tensorflow2的U

来源：萌宠菠菠乐园时间：2024-12-23 18:48

Licensed under the Apache License, Version 2.0 (the “License”);

#@title Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # https://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. 1234567891011

Image segmentation

This tutorial focuses on the task of image segmentation, using a modified U-Net.

What is image segmentation?

So far you have seen image classification, where the task of the network is to assign a label or class to an input image. However, suppose you want to know where an object is located in the image, the shape of that object, which pixel belongs to which object, etc. In this case you will want to segment the image, i.e., each pixel of the image is given a label. Thus, the task of image segmentation is to train a neural network to output a pixel-wise mask of the image. This helps in understanding the image at a much lower level, i.e., the pixel level. Image segmentation has many applications in medical imaging, self-driving cars and satellite imaging to name a few.

The dataset that will be used for this tutorial is the Oxford-IIIT Pet Dataset, created by Parkhi et al. The dataset consists of images, their corresponding labels, and pixel-wise masks. The masks are basically labels for each pixel. Each pixel is given one of three categories :

Class 1 : Pixel belonging to the pet. Class 2 : Pixel bordering the pet. Class 3 : None of the above/ Surrounding pixel.

!pip install -q git+https://github.com/tensorflow/examples.git !pip install -q -U tfds-nightly 12

import tensorflow as tf 1

from tensorflow_examples.models.pix2pix import pix2pix import tensorflow_datasets as tfds tfds.disable_progress_bar() from IPython.display import clear_output import matplotlib.pyplot as plt 1234567

Download the Oxford-IIIT Pets dataset

The dataset is already included in TensorFlow datasets, all that is needed to do is download it. The segmentation masks are included in version 3+.

dataset, info = tfds.load('oxford_iiit_pet:3.*.*', with_info=True) 1

[1mDownloading and preparing dataset oxford_iiit_pet/3.2.0 (download: 773.52 MiB, generated: 774.69 MiB, total: 1.51 GiB) to C:UsersAdministratortensorflow_datasetsoxford_iiit_pet3.2.0...[0m Shuffling and writing examples to C:UsersAdministratortensorflow_datasetsoxford_iiit_pet3.2.0.incompleteUGR50Noxford_iiit_pet-train.tfrecord Shuffling and writing examples to C:UsersAdministratortensorflow_datasetsoxford_iiit_pet3.2.0.incompleteUGR50Noxford_iiit_pet-test.tfrecord [1mDataset oxford_iiit_pet downloaded and prepared to C:UsersAdministratortensorflow_datasetsoxford_iiit_pet3.2.0. Subsequent calls will reuse this data.[0m 1234

The following code performs a simple augmentation of flipping an image. In addition, image is normalized to [0,1]. Finally, as mentioned above the pixels in the segmentation mask are labeled either {1, 2, 3}. For the sake of convenience, let’s subtract 1 from the segmentation mask, resulting in labels that are : {0, 1, 2}.

def normalize(input_image, input_mask): input_image = tf.cast(input_image, tf.float32) / 255.0 input_mask -= 1 return input_image, input_mask 1234

@tf.function def load_image_train(datapoint): input_image = tf.image.resize(datapoint['image'], (128, 128)) input_mask = tf.image.resize(data123