Visual depth affine transform

[mingyuShin]

Thank you for your great research.

Could you please let me know where I can find the code for augmenting visual depth using affine transforms with scale factor "S_y"?

Thanks

[SPengLiang]

Thanks for your interest. The scale factor is at

DID-M3D/lib/models/DID.py

Line 175 in 2475af6

scale_depth = torch.clamp((scale_box2d_masked[:,4]-scale_box2d_masked[:,2])*4, min=1.0) / \

and is used in

DID-M3D/lib/models/DID.py

Line 182 in 2475af6

vis_depth = vis_depth * scale_depth.unsqueeze(-1).unsqueeze(-1)

[mingyuShin]

Thank you for your replay.