since I'm not really experienced in CV I need your help. I would like to search but it's very difficult for me to define my search query. I hope, you won't downvote my question. And if you do, at least comment why.
In our android app we need to do document detection or better to say detection of the special kind of documents.
Let me say it should be kind of document containing title, an unique number and several checkboxes (more then 25). All of this elements are placed always at the same position.
For detection it shouldn't matter, if checkboxes are checked or not. Also the id number should be ignored.
I mean, the app should only recognize our printed document which someone holds in his / her hands at the camera.
I also want to avoid to create kind of training set / or to train our app.
What I tried:
- I tried the template algorithm. But I believe, it's a wrong way. Since I consider the whole document as a template. No live recognition was possible. But maybe I did something wrong.
OpenCV Template Matching example in Android
OpenCV Android template matching from camera
- I was able to use CraftAr Image Recognition. The main idea behind their image recognition, that we need to create a collection of images to match in the app later on.
It worked well. But I'm curious, if it's possible to find out, what algorithm do they apply for image recognition and to use it with JavaCV or OpenCV.
Are there another good libraries (commercial or free)?
Thank you in advance.
UPDATE
Since I don't have any sample yet I just created something it could look like in the future
