Scale drift and LoopClosing

[Pepper-FlavoredChewingGum]

Hi, thanks for your excellent work. After running this code, I found that the scale drift is somewhat serious, and it is difficult to detect the loopclosing. What is the reason?
Sincerely look forward to your response.

[linyicheng1]

The loopback function is disabled, and further information is needed on the scale drift. The goal of this code is currently to test the tracking performance of different matching algorithms.

[Pepper-FlavoredChewingGum]

Can you tell me how to open the loopclosing thread ? I can't find the corresponding switch, it seems that the loopclosing thread has also been started. At the same time, when I run the code on KITTI 00 sequence, the scale drift problem is serious. What is the reason?

[linyicheng1]

You can try the original ORB-slam2, which will scale drift more when there is no loopback. If you want to add loopback may be a more difficult job and may be updated in a future version. If you wish to add it yourself, you can debug from the LoopClosing::Run() function to see if each function works.

[lujunlin-HN]

Hello author, is the current version with the loopback function ready? Looking forward to your work.