Status

This module is under development. It's very raw still, the database schema is in flux, and should in no way be considered functional and/or testable. I have a roadmap I'm working towards in my spare time though, and I hope to have things more stabilized soon.

Hi Erich,

I like this initiative! This could be really cool like Google images is detecting color/shape/type of objects etc.

Currently I'm figuring out a usecase that might be common to more people.
I want to automatically categorise products into color categories based on their product image.
I really like to contribute to this module and document this usecase when I got it figured out.
As a start I want to be able to categorise the products in a few 'primary' colors (blue, yellow, red, black, white, brown, orange, green, purple), setting up taxonomy for the product content type.
A problem that may arise is that a lot of products may have white or black as a background color, and so that background color may get detected as the product color.
I cot visual_search_api_parse to give me the array of how many times a certain color appears in the image, but I am not sure how to proceed.
Can you point me in the right direction of what you think would be the right way to attack this usecase?

kind regards
Sil

I think that this is going to be quite difficult to do without code to separate shapes from the background. I don't see this as being doable anytime in the near future, so my suggestion to you is to manually tag these images with a generic color as you described, and let taxonomy handle that.

it comes to my mind that you could extract multiple colors, and if you get white or black in the bg its not a PRIORITY color. In most cases the backgroundcolor is in one of the four edges (with product imagess ONLY).