im pretty new in wikitude but tried a bit the last days. now i have a few questions how to go on.
first my general idea for my university is, that the app(which will be build with xamarin and visualstudio) should get a rest-api link to get all courses(with times and room/building) of the specific student and the app should show the courses of the day in the specific building when the smartphone "tracks" the building. "tracks" cause i dont rly know which method i should use. i tried the gps(poi) method and the Client Recognition. The Client Recognition didnt worked at all.. i tried it with pictures i made of one building and equalize perspective made things that could change (like trees) and others white - to get no points to track there. but nada :( there is the question is there a big difference in the recognition to the cloud version? i interpret the docu so, that its the same but the main computing is in the cloud.
Then the other idea with gps(poi). i use the google play service to locate the user/smartphone. but it seems like this method is mainly to use for far distances?! in my case the distances are between 2 and 30 meters. the default value here for example culling is 50km. I also have a few issues when im moving, that the pois arent updated very well.
i hope u got my problems/questions - sorry for bad english
over 1 year ago
As we currently don't have a proper large scale SLAM solution in place which would allow you to recognize/track large areas the best would be to work with geobased AR. If outside GPS should be accureate enough. As Client Recognition is optimized for 2D image recognition, it's definitely not optimal to recognize large 3D structures. In regards to the recognition itself there is no difference in Client Recognition and Cloud Recogntion in regards to quality.