lyft/nuscenes-devkit

Invalid lidar point file

DaHaiHuha opened this issue · 9 comments

During handle with the lidar point cloud, the sample(which token is
'9cb04b1a4d476fd0782431764c7b55e91c6dbcbc6197c3dab3e044f13d058011')
has a wrong dimention of lidar point file. The sensor channel is LIDAR_TOP.
When I do
scan = np.fromfile(str(file_name), dtype=np.float32) points = scan.reshape((-1, 5))[:, : cls.nbr_dims()]
the the error orrured:
ValueError: cannot reshape array of size 265728 into shape (5)

with this different lidar bin file, I can just try to handle it by this:

spec_lidar_path_1 = '/home/muzi2045/Downloads/lyft_dataset/v1.01-train/lidar/host-a011_lidar1_1233090652702363606.bin'
        if str(lidar_path) != spec_lidar_path_1:
            print("lidar_path:", str(lidar_path))
            points = np.fromfile(str(lidar_path), dtype=np.float32).reshape((-1, 5))[:, :4]
            points[:, 3] /= 255
        else:
            points = np.fromfile(str(lidar_path), dtype=np.float32).reshape((-1, 4))[:, :4]
            points[:, 3] /= 255

looks like some data wrong....
snapshot

Hi! Please make sure you work with the most recent (1.02) version of the dataset. This problem does not occur there anymore.

Is there any correct single lidar file to replace this wrong lidar data file?
Just Don't want to download the whole the dataset again...
@rkesten

Sorry, but there is no incremental update to 1.02. There are several other issues fixed as well.

I just tried to download the dataset from (https://level5.lyft.com/dataset/) again but the version of the training dataset was still v1.01. Where can I download the most recent version? @rkesten

@HH0314 good point, that link isn't up to date. This should be fixed by tomorrow, so that v1.02 will be available via that site. Sorry for the confusion!

points = scan.reshape((-1, 5))[:, : cls.nbr_dims()]
ValueError: cannot reshape array of size 265728 into shape (5)

Hey @rkesten i am also facing same issue, by downloading V1.02 can i resolve this issue?

points = scan.reshape((-1, 5))[:, : cls.nbr_dims()]
ValueError: cannot reshape array of size 265728 into shape (5)

@rkesten @explorer359
I am facing the same issue.

I cannot find the version of the dataset, but I am pretty sure it is the latest one
downloaded in July 2020 from [https://self-driving.lyft.com/level5/download/]

How can I resolve this? Any idea and comments are welcome

Thx!