search

Depth2World / NLOST

NLOST-Non-Line-of-Sight-Imaging-with-Transformer (CVPR2023)
18 stars 3 forks source link

Regarding Experimental Data #2

Open min-yang5 opened 3 months ago

min-yang5 commented 3 months ago

Hello！

I would like to learn more about the detailed settings of the experimental data; for example, the distance between the object and the scattering surface, the time resolution, the scanning range, and other key data. Could the author kindly provide me with this information?

Thank you！

Depth2World commented 3 months ago

Sry for the late. Occupied with internship. The distance varies in different scene. About 0.5m-2m. You can get the distance precisely through three views. The time resolution is 32ps for FK and NLOST real-world data. The scanning range is 2m2m ( For NLOST, 1.85m1.85m pad to 2m*2m actually ). Details about our self-built imaging system, please refer to read.me

min-yang5 commented 3 months ago

Thank you for your answer.

I would like to ask if the three traditional recovery algorithms provided in the util-pytorch file support image recovery for non-confocal TOF data?

Depth2World commented 2 months ago

No, they are all for confocal data. For non-confocal data, you can follow https://github.com/fmu2/nlos3d for the pytorch version.