The biology of lines
Teaching neural networks to sketch
From pixels to masterpieces
YOUR GOAL
Master of How AI learns to sketch the world
007skaty Artificial intelligence learns to recognize artistic boundaries from human experts
007skaty Neural Edge Detection
It traces contours through a multi-layered view, To sketch out lines the way a human artist would do.
007skaty Released in 2020, this collection of 250 annotated outdoor images teaches neural networks which edges are structurally important
007skaty Cinematic lines trace a fluffy animal's clean silhouette over a warm watercolor background.
007skaty Trace messy scribbles to master how neural networks find the edges in any drawing.
Artistic datasets: The BIPED dataset contains hundreds of outdoor photos hand-traced by professional artists just to train AI.
A rough or unfinished drawing, often depicting the basic outlines of an object. In 2017, Google researchers trained a recurrent neural network to generate these drawings using human stroke data.
Which 2019 neural edge detection algorithm uses a bi-directional cascade to capture multi-scale edges?
007skaty The transition of digital edge detection from sharp lines to artistic contours
What does the "HED" algorithm, introduced in 2015, stand for in neural edge detection?
Who developed the landmark convolutional neural network in 1989, proving CNNs could scan images?
Why does a neural network need to look at "global context" to draw a good edge map?
What are the small mathematical matrices used by CNNs to scan and filter images called?