Reference material for the SlamDynamicConfig class / Is implementation of search_map_by_projection same as ORB-SLAM2?

[Mechazo11]

Reference material for the SlamDynamicConfig class / Is implementation of search_map_by_projection same as ORB-SLAM2?

[luigifreda]

Hi,
thanks for your feedback.

1. SlamDynamicConfig is helping in robustly estimating the descriptor distance standard deviation by using MAD.
   https://github.com/luigifreda/pyslam/blob/master/utils_features.py#L133
   Moreover, see for instance: https://en.wikipedia.org/wiki/Median_absolute_deviation

2. search_map_by_projection() is working in a very standard and well-known way. As for the performance, did you have a chance to read the main README
   "
   You can use this framework as a baseline to play with local features, VO techniques and create your own (proof of concept) VO/SLAM pipeline in python. When you test it, consider that's a work in progress, a development framework written in Python, without any pretense of having state-of-the-art localization accuracy or real-time performances.
   "
   As you may know, python is not the best language to write a real-time SLAM framework. It does not allow real multi-threading capabilities: for instance, see this discussion
   https://stackoverflow.com/questions/4496680/python-threads-all-executing-on-a-single-core
   All the python threads will actually run on a single core due to GIL limitations (assuming we are using the default CPython). A better approach could be achieved by using the multiprocessing modules. Clearly, there is always space for new improvements. Feel free to open a PR for a better implementation.