Can you share what are the best practices for open world augmentation using pictures/photos taken outside from walls, post signs, monuments, street signs, posters, etc.
(note: I'm talking about image/2D recognition neither 3D recognition nor Geo Tracking/Markerless recognition).
I've made some test but it hardly work at the moment.
I'm taking picture during daytime (daylight is ok).
I believe I'm following the guidelines for the 2D trackers that you've shared in your documentation section, but more details/precision would be helpful.
I'm not sure if I have to crop the picture to augment the chance for open world recognition.
Any help would be useful.
By Open World Augmentation I mean having augmentation occurring on a part of an existing surface. For example: I take a picture of a panel at the train station --> I need to create a wtc file out of it and then display a AR Label when a user use the application built with wikitude SDK.
As I said, I tested several time this principle, but the augmentation does not occur all the time, and sometime it takes few seconds before the SDK recognizes the pattern.
I'm looking forward ways to trigger immediate augmentation by, if possible, increasing the quality of the picture taken. I've already read the documentation section "Best practice for target images", and I was looking for some best practices or technique to increase the accuracy of the recognition considering "open world" use case.
Let me know should you need more clarification.