https://github.com/thearchitector/smartclip

A contextually-aware paperclip processor that crops and scales your images to maximize aesthetic quality
https://github.com/thearchitector/smartclip

algorithm image image-processing paperclip paperclip-processor

Last synced: 25 days ago
JSON representation

A contextually-aware paperclip processor that crops and scales your images to maximize aesthetic quality

• Host: GitHub
• URL: https://github.com/thearchitector/smartclip
• Owner: thearchitector
• License: bsd-2-clause
• Created: 2017-08-20T16:31:58.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2018-05-30T15:39:06.000Z (over 6 years ago)
• Last Synced: 2024-11-14T05:34:41.047Z (about 2 months ago)
• Topics: algorithm, image, image-processing, paperclip, paperclip-processor
• Language: Ruby
• Homepage:
• Size: 17.6 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        


# Smartclip

Smartclip is a contextually-aware paperclip processor that crops and scales your images to maximize aesthetic quality.

> While initially written as a passion project for [The Glass Files](https://www.theglassfiles.com/), I've abstracted away all the hardcoded specifics so it should work generically. Images smaller than the desired thumbnail size are automatically padded with a border rather than being enlarged.

## The Algorithm 

The algorithm is fairly simple and is an adjusted a port of [smartcrop.js](https://github.com/jwagner/smartcrop.js) by Jonas Wagner:

  * Find edges

  * Find regions with a color like skin

  * Find regions high in saturation

  * Generate a set of thumbnail candidates

  * Rank candidates using an importance function to focus the detail in the center and avoid it in the edges.

  * The highest ranking candidate is selected and is processed by Paperclip

## Example

The process is super simple, and is basically identical to using any other Paperclip processor:

  ```ruby

    # Your path might be different, so adjust as necessary

    require "#{Rails.root}/lib/paperclip_processors/smartclip"

    

    ...

    styles: { thumb: { resize_width: 150, resize_height: 150, processors: [:smartclip] } }

  ```

## Future Plans

  * Wrap within gem for easy installation

  * Optimize pixel iteration to reduce number of loops. Can it be done without screwing with saliency patterns?

  * Port and integrate face detection algorithm

  * Completely rewrite to favor deep neural networks? Benefits from more complicated criteria, and can be trained from professional data rather than guesses

  * _Extract optimzation logic, along with what I've learned, to create separate image optmization processor?_

## Found Bugs?

If you found a bug, submit an issue. If you have some optimization insight, submit a Pull Request.