Computer Vision with Neural Networks

Bring the missing detail back

A journey through u-net with Botie.

Computer Vision with Neural Networks · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

How Do We Identify an Object Without Losing Its Boundary?

Part 1 of 3

Knowing that a microscope image contains cells is not enough if we need to measure each cell. We need to identify which pixels belong to the object. Reducing an image through a classification network helps collect context, but it can discard the fine detail needed to draw a boundary.

Sources for this history

Ronneberger, Fischer and Brox (2015), U-Net: Convolutional Networks for Biomedical Image Segmentation

← Previous lesson: Convolutional Networks Read the full lesson and derivations