Chinese and Japanese Handwriting recognition. Port of tegaki wagomu engine
To use this package, run the following command in your project's root directory:
slightly modified wagomu engine from tegaki project ported to D.
Grab a wagomu model from https://tegaki.github.io/ and load and use it using:
import wagomu; Recognizer r = Recognizer(2); // 2 = how many strokes the drawn characters can be off the actual character r.load("/usr/share/tegaki/models/wagomu/joyo-kanji.model"); // put your model path in here // this model uses a canvas of 1000x1000, check the xml file with your model for info Character ch = Character(20, 1); // point count, stroke count for (int x = 0; x < 10; x++) ch.points[x] = [x * 80 + 100, 500, 0, 0]; // generate straight line from x 100 to 900, y = 500 // you can use user input here // argument 3 and 4 are unused in this model, they might be associated with input like pressure or thickness, depending on the model // points must be in stroke order, density should be roughly 50 pixels apart each point for this model (10 points per straight line) // higher density still works but the higher the density is, the more likely it is going to compare with characters with lots of strokes auto res = r.recognize(ch, 5); // returns up to 5 suggestions for this character, best suggestion first assert(res.unicode == '一');
- 1.0.0 released 2 years ago
- Copyright © 2017, webfreak