The new constraint about pose is not useful?

nianticlabs / monodepth2

[ICCV 2019] Monocular depth estimation from a single image

Other

4.16k stars 962 forks source link

The new constraint about pose is not useful? #492

Open Zhangwenyao1 opened 8 months ago

Zhangwenyao1 commented 8 months ago

Thanks for your great work! I found it is not useful if I employ the constraints ( MSE(T(-1, 1}, T(1,0) * T(0,-1))) as one type loss, and the abs would decrease, So can you give me some advice?