Awesome

Vision

Cross framework, cross platform computer vision library for Haxe.

• Structure
• Tutorials
• Versioning
• About
• Roadmap
• Changelog
• Define List

Structure

This Library is organized into 5 subfolders, and one important file:

• ds - contains Vision's data structures & types. Some functions and Features require different return types which aren't basic - such as lines, rays or points. Those can be found inside this folder.
• tools - contains classes that extend the functionality of certain aspects of the library. For example, MathTools is a superset of the Math class from the standard library, and also adds many features related to finding data related to lines, rays and points.
• algorithms - contains every complex algorithm this library uses. They're all contained in one folder to help you learn from them :). If you feel like some algorithms are lacking/have unclear documentation, please open an issue about it
• exceptions - contains the errors usually thrown by Vision. Notice - those errors are present just to let you know if you did something unexpected. If you want to get rid of them, define vision_quiet.
• helpers - contains some helper classes that may be used by Vision. Theres nothing preventing you from using them yourself though...
• Vision.hx - contains all image manipulation/computer vision methods. After 1.0.0, this class should always remain backwards compatible. It is always recommended to use the methods from this class instead for using the ones in the wanted algorithm's class, since the ones in Vision are much less likely to break compatibility, are usually simpler to use, and have much better documentation.

Tutorials

For more tutorials (or further details about the tutorials here) check out the links below:

• Vision Tutorials On The Haxe Learning Center

Creating an image

Create a blank image for drawing:

var image = new Image(200, 300, Color.WHITE); //creates a new, 200x300 image with a white background

Drawing:

image.setPixel(10, 35, 0xff3204);
image.drawLine(1,1, 34, 67, Color.LIME);

Create an image from a URL:

var image:Image;
ImageTools.loadFromFile("https://upload.wikimedia.org/wikipedia/commons/thumb/f/f0/Valve_original_%281%29.PNG/300px-Valve_original_%281%29.PNG",
    data -> {
        image = data;
    }
);

Add that image to the screen (available on platforms & libraries that support drawing on screen):

using vision.tools.ImageTools;

//...

image.addToScreen(0, 0); //alternatively, you can do - ImageTools.addToScreen(image, 0, 0)

Basic Image Manipulation

Vision.grayscale(image);
Vision.sharpen(image);
Vision.convolve(image, BoxBlur);

OR

using vision.Vision;

image.grayscale();
image.sharpen();
image.convolve(BoxBlur);

Implicit Conversions From/To Framework Specific Graphical Types

import openfl.display.BitmapData;
import vision.ds.Image;

var future = BitmapData.loadFromFile("https://upload.wikimedia.org/wikipedia/commons/thumb/f/f0/Valve_original_%281%29.PNG/300px-Valve_original_%281%29.PNG");
future.onComplete(bitmapData -> {
	var image:Image = bitmapData; // converts the given openfl.display.BitmapData to vision.ds.Image
   // The conversions are lossless, so you never lose data.
	addChild(new Bitmap(image)); //converts back to a BitmapData.
});

Using MathTools

import vision.tools.MathTools;
using vision.tools.MathTools;

var ray = new Ray2D({x: 0, y: 0}, 1); //f(x) = x
trace(ray.distanceFromRayToPoint2D({1, 0}) /* function from MathTools */); // sqrt(2) / 2
trace(MathTools.median([[12, 123, 321, 4, 333], [12, 123, 145, 321, 4], [12, 123, 264, 321, 4], [12, 123, 234, 321, 4]].flatten())); //123;
//you can even chain calls
trace([[[[[1, 2, 3, 4], [5, 6, 7]]]]].flatten().flatten().flatten().flatten().median()); // 4

Versioning

Vision's versioning will try to follow semver's specifications as accuratly as possible. Here is a list of what can change across version types:

✔️: This type of modification is allowed
❌: This type of modification is disallowed
❓: This type of modification may occur from time to time

About

Vision is a cross framework, cross platform Computer Vision & Image Manipulation library for Haxe.
This library exists to provide CV & IM capabilities to Haxe projects, naturally:

What Does That Mean?

Basically, if you're using HaxeFlixel for example, you should be able to easily use an FlxSprite as an image to process, or if you're using OpenFL you should be able to use bitmaps and sprites to process.

Why Does This Library Exist?

One of the problems i have with Haxe is the lack of non-game utilities and libraries. my goal is to provide those utilities, and make app development in Haxe easier.

Contributions Are Welcomed.

I'm not the best at everything I do, especially in fields like computer vision, with which im not the most familiar. That means there are bound to be mistakes/bugs. Because of that, every issue, pull request or code review is welcomed. Help me make this library bigger and better for everyone :)

About Documentation

One of the things I can't stand with libraries similar to this one is the lack of docs, and unclear, obfuscated code.

If you see some code that you think is not understandable, or some place that lacks documentation, or even inaccurate/unclear documentation, please open an issue about it, and I'd try to resolve it asap ;).

Define List: