vcciv/blvd

Building A Large-scale 5D Semantics Benchmark for Autonomous Driving

blvd's Issues

Download to linux

Dear BLVD-group,
Is it possible to download the data to linux and without a baidu account? As we tried to download the data it offered only an .exe to download. Which labels should we click on to access the data after logging in?

The google drive link need permission request

Hi,

Thank you for your great dataset. Could you please share with me the permission to download data from google drive link ? Thank you so much.

How to transform the yaw angle from velodyne to cam coordinate?

Thank you for your dataset!
The script your provided in "utils.py" can calculate yaw angle. But maybe is it in the velodyne coordinate? How to transform this yaw angle to camera coordinate, just like KITTI dataset?
Thank you!

How can I get the data set

I sent a message but didn't get a reply.How can I get the data set? Thanks

show targets in lidar coordinate

Hi, thanks very much for your work and it's useful for most people.
But there are some problems when using your dataset.
I am wondering why the boxes3d dimensions of vehicle in label are small , compared to (l w h : 3.9,1.6,1.56 in kitti). It doesn't seem to contain a full target object.
I want to show lidar with ground truth boxes in lidar coords, the below is output of compute_box_corners_3d. The target box is not in the right direction. The length of x aixs is 0.94. the minimum and maximum are not the same as that in label.

Can you kindly tell me what's wrong and show us an example of drawing lidar with gt boxes? Thank you very much.

Label.txt
1 3 5 2 11 9.915727 -3.910213 2.288099 0.942343 3.432148 0.000000 -2.193600 -0.496231

Calib.print_object()
3d bbox h,w,l: 1.697369, 0.942343, 2.288099
3d bbox location, ry: (9.915727, -3.910213, -1.344915), (1.000000,0.000000)

Results of compute_box_corners_3d
box3d_velo: [[10.3868985 -3.910213 -2.488965 ]
[ 9.4445555 -3.910213 -2.488965 ]
[ 9.4445555 -3.910213 -0.200866 ]
[10.3868985 -3.910213 -0.200866 ]
[10.3868985 -5.607582 -2.488965 ]
[ 9.4445555 -5.607582 -2.488965 ]
[ 9.4445555 -5.607582 -0.200866 ]
[10.3868985 -5.607582 -0.200866 ]]

License Information

Dear,
Please could you provide the License for the dataset and repository.

Is 2D label availabel?

Thanks for sharing your great work!
I want to know is there any 2D label in the label information besides 3D bbox?

do this dataset has poses from lidar2world?

Baidu link dead

Looks like Baidu link should be updated.