Overview
AI, Data Science & Cloud Certificates from Google, IBM & Meta — 40% Off
One plan covers every Professional Certificate on Coursera. 40% off Coursera Plus Annual.
Unlock All Certificates
This 42-minute tutorial demonstrates how to achieve exceptional accuracy in water segmentation from satellite imagery using EfficientNetB0. Learn to build a U-Net architecture with EfficientNetB0 as the backbone, create an efficient data preprocessing pipeline, train the model on a water segmentation dataset, and perform inference on both single and multiple test images. The tutorial is structured in clear chapters, beginning with an introduction and demo, followed by installation instructions, model building, data preparation, training, and inference techniques. Access the complete code through the provided Ko-fi link, and explore additional computer vision and image segmentation tutorials on the creator's blog and YouTube playlists. Perfect for computer vision enthusiasts looking to implement advanced segmentation techniques with state-of-the-art accuracy.
Syllabus
00:00 Introduction and Demo
02:38 Installation
08:14 Build the Efficientnetb0 U-net model
13:28 Prepare the data
24:28 Build the model + training
27:27 Run inference on single image
34:29 Run inference on multiple test images
Taught by
Eran Feit