Dataset and running code.

[edgr-coder]

Your project is very interesting. May you send your dataset and running code to me? cumtchenchang@163.com. I also have another question, is your code is suitable for OKVIS? For example, in okvise, just directly replace your code with ceres-solver, the project can still run successfully.

[panxkun]

Thank you for your interest. Since I haven’t worked on the relevant branch for a while, I need to evaluate whether the current codebase is ready for open-sourcing. However, you can refer to XRSLAM, which already integrates dataset loaders and a ceres-solver-based optimization interface.

The current repository provides a custom implementation of the least squares solver based on the same interface, making it essentially a plug-and-play module. So if you urgently need this implementation, you can port tinysolver into XRSLAM yourself, with only minimal interface adaptation required. If you are reasonably familiar with how to use ceres-solver and have a well understanding of basic cpp programming, I believe it should only take a day or two to complete this task.

Regarding the OKVIS algorithm you mentioned, I haven’t studied it in detail. But since you said it’s implemented based on ceres-solver, I believe it’s feasible. You would just need to adapt the evaluate function to make it compatible with the ceres-solver-style interface.