Deep learning methods for RTAB-Map

[matteo.scucchia]

Hi,
i'm a Computer Science master student at Bologna University, and for my master degree thesis i have to apply deep learning methods to Visual SLAM.

I'm new in RTAB-Map and i started using RTAB-Map with an Intel realsense D435 camera for detecting loop closure.

My work will be use deep learning method (CNN) to extract features from images for the loop closure detection task. I have to apply lifelong learning methods too in order to obtain a lifelong learning of the environments.

My question is if RTAB-Map can be a good starting point to apply this methods and if there are flexible ways to add this kind of functionalities or not.  

Thanks in advance.