Closed nesquik011 closed 4 weeks ago
Hello,
We only use monocular (RGB only) as input. All dataloaders only return a timestamp and RGB image. For a live camera dataloader, please see our RealSense example https://github.com/edexheim/como/blob/da616363454591085c3f02715d7d9df04388a19f/como/data/RealsenseDataset.py#L109 Note that we do not use the depth stream at all.
Best, Eric
@edexheim how did you manage to get good depth ? i am using stereo camera with slam , and i cant reach 50 % of your depth results i am using SGBM so is that deep learning thing or what is that ?
Hello,
Yes, our system uses a combination of a learned prior combined with multi-view optimization. Please see our papers linked for more details on the specifics.
Best, Eric
kindly what type of camera is it ? normal mono camera ? or depth camera ?