Deep Extreme Cut

www.vision.ee.ethz.ch

A great paper  which explores the use of extreme points in an object (left-most, right-most, top, bottom pixels) as input to obtain precise object segmentation for images and videos. We do so by adding an extra channel to the image in the input of a convolutional neural network (CNN), which contains a Gaussian centered in each of the extreme points. The CNN learns to transform this information into a segmentation of an object that matches those extreme points.

Read more...
Linkedin

Want to receive more content like this in your inbox?