Accuracy of VINS-Mono on the Euroc Dataset

[vladniculescu]

Hello everyone,

Has anyone tried to reproduce the experiments from the VINS-Mono paper to compute the ATE on the five EuRoC datasets? I consistently obtain results in the range of 0.2–0.6 m, whereas the paper reports 0.15–0.3 m. I’m wondering if I might be doing something wrong in my evaluation pipeline.

Additionally, does anyone know whether the exact same configuration was used for all EuRoC sequences in the original benchmarks, or if certain parameters were tuned per dataset?

Any insight would be greatly appreciated!

[heroamd]

I have the same accuracy queston. Have you have solved? @vladniculescu