Discrepancy in RMSE Results on EuRoC Using evo Evaluation

[iuenao]

I tested this library on the EuRoC dataset and evaluated the trajectories using the evo tools (https://github.com/MichaelGrupp/evo).

I noticed that the RMSE values I obtain are noticeably different from those reported in the original paper. For example, for the V1_03_difficult sequence, the VINS-Mono RMSE without loop closure is reported as 0.18 m in the paper, but my evaluation gives around 0.25 m.

Could you please advise what might cause this discrepancy?