Visual SLAM Toolkit tracking issue
Hi,
I used the Visual SLAM Tool recently (which I got it from Play Store). But the only issue I faced is after I scan the model from all angles and save the map the application identifies the scanned area only if the position and angle matches the initial position and angle at which I started scanning.
Lets Assume this scenario, I start scanning a model with my device pointing north and I scan the whole model at all angles and I get the yellow indicators for the mapped angles and create/save a 3D map of it. Then I load the same map and test it out on the same model but this time my device facing west (please note that the model has been mapped from these angles too) and it wont recognize the model. It only works when my device points west which is the initial frame's position and angle. To make it work, my device angle and position should be equal to the inital device angle and position while scanning. After that everything works as expected.
I faced this issue some time back when I tried Object Tracker. But due to this issue I had to look out for alternate options. I am planning to integrate Object Tracker into my existing project, but only if this issue has a solution. I tried this on Android and Windows.
Expected Results: If I have scanned a model from all angles then it should identify the object from all those angles.and positions.
Can this issue be fixed? Or is there a workaround for it?
Thanks in advance.
Thanks for your interest in MAXST SDK.
I saw the contents of the story well.
The bottom line is that if you scan the model at all angles, the object is identified at all angles.
I also learned that after learning an object from another angle, the position of the object changes or the angle of the object changes, and then it is recognized even if it is tested again.
There are many reasons why you may not recognize objects when you move objects at different angles.
The target feature is not detected much, the light is influenced by the lighting, and a lot of characteristic points are generated in the surrounding environment.
If you still can not recognize the object, you should change the target and change the lighting. The lighting is good for the surface light source.