tum-vision / lsd_slam

LSD-SLAM
GNU General Public License v3.0
2.58k stars 1.23k forks source link

some questions about omni-lsdslam #318

Open baozeng90 opened 5 years ago

baozeng90 commented 5 years ago

when I download data from https://vision.in.tum.de/data/datasets/omni-lsdslam.I do not know the parameters in Calibration File:Omnidirectional:calibrationOrig.yaml like intrinsics: [2.0616467,442.843,442.66,320.65,253.91].what it means for intrinsics[0] [1] [2] [3] [4]?

EnriqueSolarte commented 5 years ago

So far, the first parameter corresponds to epsilon (shifting value). The value for this shifting distance was estimated to 2.06, the paper said that. But, I don't know how to use this parameter in the lsd_slam. The rest of the parameters corresponds to fx, fy, cx, cy.

Have you already run lsd-slam for omnidirectional cameras?

fangxu622 commented 5 years ago

I also have this problem. how to run lsd-slam with omni camera ~~do you resolve this? Who can give me some help? ~ thanks very much @baozeng90 @EnriqueSolarte @christiankerl @pierriko @efernandez @

fangxu622 commented 5 years ago

@SummerIcequeen @ffurrer @lesire

fangxu622 commented 5 years ago

@jacksonkr

Dvad commented 5 years ago

Hi, The omnidirectional version of LSD-SLAM has never been released publicly. This repo is only for the regular pinhole model.

EnriqueSolarte commented 5 years ago

Hi, Thank @Dvad. Yes, after diving in the code for a while I realized that. I am implementing omnidirectional based on this version and the paper details. I already introduced it the unified model, but I am struggling on the stereo matching and variance propagation. Any idea guys? Thank you so much in advances.