Hi,

Check out our new tutorial by Margaret Maynard-Reid, U-Net Image Segmentation in Keras, on how to train a U-Net to segment pet images in TensorFlow 2 / Keras. 

Image

Image segmentation is an important computer vision task with wide applications such as medical imaging, clothes segmentation, flooding maps, self-driving cars, etc. 

U-Net is a great fundamental architecture for learning semantic segmentation on images, and you may find it in other models such as the generator of a Generative Adversarial Network (GAN).

After a brief intro to image segmentation and U-Net architecture, we will walk through the code implementation in the Colab notebook. You will learn how to load the Oxford-IIIT pet data with the TensorFlow dataset and how to train an image segmentation U-Net model from scratch. We will create the U-Net with Keras Functional API and visualize the U-shaped architecture with skip connections. After the training, we will use the trained model to make predictions on the test dataset. 

Click here to read the full tutorial

This lesson is part of PyImageSearch University under Deep Learning 105 — Hands-on Experience with CNNs, and the direct link is here.

Have fun learning! 
 

PyImageSearch Team 

P.S. PyImageSearch University helps you master computer vision, deep learning, and OpenCV with content from experts. It has 35+ courses (updated regularly), 39+ hours of on-demand video, and access to centralized code repos for all 500+ tutorials on the PyImageSearch blog. In addition, you can follow each lesson easily with code in Colab notebooks in a web browser on any OS.

Click here to join PyImageSearch University