Quan

Semantic segmentation labels on stereo frames

3

I am interested in semantic segmentation using stereo cameras. If I understand correctly, the semantic annotation labels are on the monocular RGB camera (pylon), but not the left or right...

Why are all calibration matrices identity transforms in calib.txt?

1

Thanks for providing such a large dataset. I checked a few `calib.txt` ( `calib.txt` -- calibration of velodyne vs. camera. needed for projection of point cloud into camera. ) under...

Is this a bug in https://github.com/kujason/wavedata/blob/c4b5aabd9eb3b74fad777349f75161032d3460fa/wavedata/tools/core/voxel_grid_2d.py#L67-L73 The height within a voxel is lost when it discretizes the points by voxel size. Shouldn't it be the following to pick the highest...

swin2_tiny failed to run forward(): RuntimeError: unflatten: Provided sizes [64, 64] don't multiply up to the size of dim 2 (64) in the input tensor.

1

I tried with ```python model = DPTDepthModel( path=None, backbone="swin2t16_256", non_negative=True, ) ``` During inference at https://github.com/isl-org/MiDaS/blob/bdc4ed64c095e026dc0a2f17cabb14d58263decb/midas/backbones/utils.py#L72 it gave the error ``` RuntimeError: unflatten: Provided sizes [64, 64] don't multiply up...