DAIR数据集转成KITTI数据集后车头位置有问题

[Jarvisxize]

DAIR的路侧数据转成KITTI数据后通过可视化展示后，发现车头的位置顺时针偏了90度，即车头都变成了车的右侧，这个问题如何解决

[Ignite616]

我也遇到了相同的问题,请问您怎么解决的

[Ignite616]

您好，我将gen_kitti/label_json2kitti.py中write_kitti_in_txt()函数的
i15 = str(-eval(item["rotation"])) 修改为 i15 = str(-eval(item["rotation"]) - np.pi/2)
i9, i11, i10 = str(item["3d_dimensions"]["h"]),str(item["3d_dimensions"]["w"]),str(item["3d_dimensions"]["l"]) 修改为
i9, i11, i10 = str(item["3d_dimensions"]["h"]),str(item["3d_dimensions"]["l"]),str(item["3d_dimensions"]["w"])
这两处修改，解决了这个问题。

[yaobaishen]

in my case, the converted kitti format label is NOT on the same horizontal plane with the lidar pointcloud, which leads to the label unusable, anyone see this issue?

[yaobaishen]

in my case, the converted kitti format label is NOT on the same horizontal plane with the lidar pointcloud, which leads to the label unusable, anyone see this issue?

please ignore this... it's a bug in the visualization tool, not the kitti converter.
I have filed an issue in the visualization tool retpo: zzzxxxttt/simple_kitti_visualization#2