Zhengyang Feng

> I have Pytorch 1.10.2. > >>> import torch > >>> print(torch.__version__) > 1.10.2+cu102 > try 1.6 or 1.8 instead. And of course the corresponding torchvision version.

@Wakarox Yes it should be mmcv, but I never encountered this before. Maybe try find it in the [mmcv repo](https://github.com/open-mmlab/mmcv)? However, are you installing things directly to your system (.local)?...

@solidexu In lane detection, we do not use validation for possible early stopping. Since almost everyone don't use online validation in this field, and in earlier formulation of this framework,...

@JiweiMaster 2D车道线检测输出的相当于是 uv 坐标，如果有相机参数和深度信息，可以投影到3D。 这里讲了基本的形式：https://zhuanlan.zhihu.com/p/144307108 实际的相机参数矩阵自由度可能更高，比如内参还可以有axis skew，这取决于你的相机模型。

@AirSmoke031 If you load a tusimple trained pt file while using a CULane config pipeline (i.e. just modify the model so it can load the tusimple trained pt file). It...

As for label format convertor, we don't have that yet.

Anyone providing some kind of label conversion script, is encouraged to make a pull request and add them in `/tools/`

@RyanHangZhou Do you mean something like visualizing a line (defined by sample points), on a given image?

tools/vis/lane_img_dir.py can visualize CULane format line txts on a folder of images.

you'll need to install it via `pip install importmagician`