if without ground truth，如何进行误差评估呢？

[debuleilei]

周老师你好！测试了您的这个库，结果很惊艳，可是当我用来进行人脸建模的匹配上时，多帧融合之后，还是会引入不小的误差，所以想请教一下，有没有类似重投影误差这样一种评估的方法，来筛选FGR得到的位姿结果呢？Evaluation，需要输入gt.lg, 如果没有怎么办？

[qianyizh]

I think you can run ICP based on the FGR result and evaluate on that. You can check our latest library Open3D. www.open3d.org It has much more registration algorithms and related evaluation. Hope this helps. Qianyi

…

On Sat, Sep 29, 2018 at 7:52 PM 18500864609 ***@***.***> wrote: 周老师你好！测试了您的这个库，结果很惊艳，可是当我用来进行人脸建模的匹配上时，多帧融合之后，还是会引入不小的误差，所以想请教一下，有没有类似重投影误差这样一种评估的方法，来筛选FGR得到的位姿结果呢？Evaluation，需要输入gt.lg, 如果没有怎么办？ — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#15>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFmmEVT18lB_EeW7Tzw2cZbD0T44bXILks5ugDIHgaJpZM4XAyAn> .