deeplearning

deeplearning / ArtsyNetworks / 0.2.10

README.md

Table of Contents

Drawing

artistic style transfer after 10 iterations

Introduction

This algorithm takes a photo or image sample, along with a style example from a particular artist or art form (van gogh, cubism, etc) and attempts to create a stylized version of the image sample.

Note:The first call to this algorithm will take a bit longer than sequential calls to due algorithm initialization. All following calls will be significantly faster.

Note: This algorithm takes longer than the default minimum timeout, please define a custom timeout greater than 5 minutes or else the result will be undefined.

I/O

Input

{  
   "source" : "data://zeryx/ArtsyNN/headshot.jpg",
   "style" : "https://www.ibiblio.org/wm/paint/auth/caravaggio/st-thomas.jpg",
   "iterations" : 3,
   "output_name" : "output.png"
}
  • source - (required) - the source image to classify, this must be in one of the following formats: data URI, web http/https remote file http://.. or https://..., binary image or a base64 encoded JPEG String.
  • style - (required) - the style image to reference for artistic inspiration, this must be in one of the following formats: data URI, web http/https remote file http://.. or https://..., binary image or a base64 encoded JPEG String.
  • iterations - (optional)* - the number of iterations for the algorithm to "pass" over the generated image, each consecutive pass improves the quality but dramatically increases compute time. (a range between 1-4 is great for fast results)

Output

"output.png"
  • output - the resultant output image generated from the model, the image is stored within the algorithms temp data collection for convenience.

Credits

This algorithm was originally sourced from the ArtsyNetworks github, and modified to work within algorithmia. The original work is from the Neural Algorithm for Artistic Style paper.