Deep Learning Style Transformations in Ruby
kojix2
Posted on September 3, 2019
Want to do DeepLearning inference in Ruby?
Yes, you can!
Requirement
ONNXRuntime and its Ruby binding created by Andrew Kane.
Liquid error: internal
st.rb
require 'magro'
require 'onnxruntime'
model = OnnxRuntime::Model.new('')
input = Magro::IO.imread(ARGV[0])
.transpose(2, 0, 1)
.expand_dims(0)
.to_a
output = model.predict(input1: input)
result = Numo::Int32[*output['output1'][0]]
.clip(0, 255)
.transpose(1, 2, 0)
.cast_to(Numo::UInt8)
Magro::IO.imsave(ARGV[1], result)
Run
Input image size should be a multiple of 4
ruby st.rb in.jpg out.jpg
Generated images
https://commons.wikimedia.org/wiki/File:Arimatsushibori.JPG 1
Beautiful.
Other posts
-
Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0) ↩
💖 💪 🙅 🚩
kojix2
Posted on September 3, 2019
Join Our Newsletter. No Spam, Only the good stuff.
Sign up to receive the latest update from our blog.