ukyvision

ukyvision / deephorizon / 0.1.0

README.md

1. Introduction

A CNN-based horizon line estimation technique, introduced in the following publication:

Horizon Lines in the Wild (Scott Workman, Menghua Zhai, Nathan Jacobs), In British Machine Vision Conference (BMVC), 2016. pdf
@inproceedings{workman2016hlw,
  title = {Horizon Lines in the Wild},
  author = {Workman, Scott and Zhai, Menghua and Jacobs, Nathan},
  booktitle = {British Machine Vision Conference},
  year = {2016}
}

Input:

  • (Required) Image Data API Url, Web (http/https) Url, binary image or a base64 encoded image.

Output:

  • The horizon line location in image space.

2. Examples

Example 1.

  • Parameter 1: Data API Url
{
  "image": "data://ukyvision/deephorizon/ecd_001.jpg"
}

Output

{
  "left": [-576, 26.96405317924003],
  "right": [576, 16.142339480065232]
}

Example 2.

  • Parameter 1: HTTP Url
{
  "image": "https://upload.wikimedia.org/wikipedia/commons/thumb/9/90/HFX_Airport_4.jpg/1280px-HFX_Airport_4.jpg"
}

Output:

{
  "left": [-640, 294.98985200147615],
  "right": [640, 292.69899981705186]
}

Example 3.

  • Parameter 1: Base64 image
{
    "image": "data:image/png;base64....",
}

Output;

{
  "left": [...], 
"right": [...] }

3. Credits

For more information, please refer to: http://hlw.csr.uky.edu

Demo image(s) were retrieved from:

https://en.wikipedia.org/wiki/Airport#/media/File:HFX_Airport_4.jpg