Submitted anonymously.

Submission data

Full nameDeepMVS_CX
DescriptionWe use DeepMVS to estimate disparity maps and then convert them to depth maps. Depth maps are fused into point clouds via the fusion method provided by COLMAP. As DeepMVS does not estimate normal maps, we generate fake normal maps and invalidate their contribution in the fusion step by parameter tuning.
We did not record the run-times so they are all set to 1.
Reason for invalid results for some of the datasets is that they were not generated before our submission, but not that DeepMVS failed in these datasets.
ParametersFor disparity map estimation via DeepMVS, we set width = 2000 and height = 1332 due to limited memory.
For depth map fusion, we use default parameters provided by COLMAP except for max_normal_error = 100.00.
Programming language(s)Python
HardwareGTX 2080Ti
Submission creation date26 Jun, 2019
Last edited27 Jun, 2019

High-res multi-view results



Infoallhigh-res
multi-view
indooroutdoorbotani.boulde.bridgedoorexhibi.lectur.living.loungeobserv.old co.statueterrac.
copyleft27.4040.1726.9024.5917.4423.9548.619.7726.4818.0624.4728.83

Low-res many-view results



Infoalllow-res
many-view
indooroutdoordelivery areaelectroforestplaygroundterrains
No results yet.

Low-res two-view results



Infoalldeliv. area 1ldeliv. area 1sdeliv. area 2ldeliv. area 2sdeliv. area 3ldeliv. area 3select. 1lelect. 1select. 2lelect. 2select. 3lelect. 3sfacade 1sforest 1sforest 2splayg. 1lplayg. 1splayg. 2lplayg. 2splayg. 3lplayg. 3sterra. 1sterra. 2sterra. 1lterra. 1sterra. 2lterra. 2s
No results yet.

SLAM results



allboxesboxes darkbuddhacables 4cables 5desk 1desk 2desk changing 2desk dark 1desk dark 2desk global light changesdesk ir lightdinodroneforeground occlusionhelmetkidnap 2lamplarge loop 2large loop 3large non loopmotion 2motion 3motion 4planar 1reflective 2scale changetable 1table 2table 5table 6table global light changestable local light changestable scenetrashbin
MethodInfo
No results yet.