search

m-dayani / EORB_SLAM

An event-based ORB-SLAM algorithm
GNU General Public License v3.0
3 stars 0 forks source link

What parameters do I need to adjust when I run my own event-based dataset? #2

Open Sarah-happyeveryday opened 5 months ago

Sarah-happyeveryday commented 5 months ago

[First of all, the following is the content translated by the translator, if there is not quite correct grammar, please forgive me.]

Hello, thanks for your excellent work! Here are my questions:

1、I have successfully run the code with some public datasets. Now I want to run it with my own recorded dataset (with monocular event-based camera only, without imu data), what parameters do I have to change?

2、I see the DS.seq.target section, which I don't understand, how do I set this value if I want to run my own data?

3、Also at Event Parameters, for parameter Event.data.l1ChunkSize, given a number of values corresponding to the public datasets, what does this parameter mean? How can I modify it to fit my own data?

4、In addition, if my image sequence is produced using event integration, rather than frame camera, can it be well constructed the map?

Looking forward to your reply!

Sarah-happyeveryday commented 5 months ago

One more thing, there are a lot of cpp files that look similar, right? (like EventFrame.cpp and EventFrame-1.cpp) I wonder, what files should I pay more attention to if I want to learn event-based SLAM with this code? Because there are so many cpp files……

Sarah-happyeveryday commented 5 months ago

Hi, now I have addressed the second and fifth points above. And I evaluate the dataset with the evo tool, like evo_rpe tum ……groundtruth.txt. I don't know how well it evaluates event data because I don't understand the .py files in the evaluation folder. I'm so sorry, but I'm really a novice. Could you please tell me which .py files to use and which parameters to adjust? Thank you~