Submitted by jang hyoseok.

Submission data

Full namePatchmatchNet: Learned Multi-View Patchmatch Stereo
DescriptionLearning-based cascade multi-view stereo using PatchMatch optimization. Iteratively estimates depth maps using learned matching costs and adaptive propagation.
Parameters* Number of views: 7
* Image max dimension: 2688
* Geometric mask threshold: 2
* Photometric threshold: 0.6
* Pretrained checkpoint: params_000007.ckpt
Publication titlePatchmatchNet: Learned Multi-View Patchmatch Stereo
Publication authorsFangjinhua Wang, Silvano Galliani, Christoph Vogel, Pablo Speciale, Marc Pollefeys
Publication venueCVPR 2021
Publication URLhttps://arxiv.org/abs/2012.01411
Programming language(s)Python with CUDA
HardwareGeForce RTX 3060
Source code or download URLhttps://github.com/FangjinhuaWang/PatchmatchNet
Submission creation date15 Jun, 2026
Last edited15 Jun, 2026

High-res multi-view results



Infoallhigh-res
multi-view
indooroutdoorbotani.boulde.bridgedoorexhibi.lectur.living.loungeobserv.old co.statueterrac.
copyleft91.9796.2098.59

Low-res many-view results



Infoalllow-res
many-view
indooroutdoorlakesidesand boxstorage roomstorage room 2tunnel
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.