1) the scan area is small, in the video the user can scan all room, how to do in unity??..
You could look at the Room Capture Sample within the SDK for examples on how to an entire room rather than just a small section. Specifically, you should look at the setupMapper function in viewController+SLAM.mm, and renderScanningVolumeFeedbackOverlayWithDepthFrame and renderSceneWithDepthFrame in viewController+OpenGL.mm
I would like to let you know that this even has limitations, as the largest limiting factor for the size of your scan is the memory (RAM, not storage) on your iPad, which is typically 1, 2, or 4 GB. Some of the apps that we have developed can scan an area that is roughly 400 sqft in dimension.
2) How to save mesh in application/Unity and position the point 3d in precision position?
Could you please explain this question a little more, as I don't quite exactly know what you mean? Thanks!
3) Can I recognize the Object at a later time ??
Although the Structure SDK does not do object recognition, you could use other libraries outside of our SDK to do so. You can scan a a room and then place some sort of marker on an object (a portion of the 3D Mesh) to be detected or "seen" later on by your program, but the scanner/SDK is not going to know what an object is, unless you have a ton of data of other scanned objects.
For example, I can scan a room and mark a chair with some sort of marker to let the app know that that is a chair. I can't design an app that will automatically know that a chair has been scanned unless you have a ton of other data to help the app out.
Here are a couple of things I have found that could help you discover how to do this: