Skip to main content Link Search Menu Expand Document (external link)
ORML

Table of contents

  1. What can I do with it?
  2. How do I use it?
  3. Credits and references

orml-super-resolution

Image up-scaling based on FALSR super resolution neural network

What can I do with it?

orml-super-resolution can be used to up-scale images. The FALSR network usually generates sharper images than bilinear upscaling.

upscaler-01.png

How do I use it?

To load the up-scaler:

val upscaler = ImageUpscaler.load()

To upscale a single image:

val image = loadImage("<some-image>")
val upscaled = upscaler.upscale(image)

DemoUpscale.kt demonstrates the full process of upscaling images.

Credits and references

orml-super-resolution uses a pretrained network from https://github.com/xiaomi-automl/FALSR, to be exact it uses the FALSR-A variant.

@article{chu2019fast,
  title={Fast, accurate and lightweight super-resolution with neural architecture search},
  author={Chu, Xiangxiang and Zhang, Bo and Ma, Hailong and Xu, Ruijun and Li, Jixiang and Li, Qingyuan},
  journal={arXiv preprint arXiv:1901.07261},
  year={2019}
}