CLRNet

CLRNet copied to clipboard

Are images normalized during training/inference?

1

I can see that RGB/BGR (?) normalization is defined here https://github.com/Turoad/CLRNet/blob/d0fef907aae323ef37da0c176c5df35a0565e58d/clrnet/datasets/process/transforms.py#L276 and that image_norm information is defined in the configs: https://github.com/Turoad/CLRNet/blob/7269e9d1c1c650343b6c7febb8e764be538b1aed/configs/clrnet/clr_dla34_culane.py#L44 but as far as I can tell, grepping through...

LukeAI

out of index error when testing with CULane dataset

3

Tried on both culane_res18 and culane_res34 both two models got this out of index error in clr_head 使用预训练culane模型进行验证出现out of index 错误 pytorch=1.8 torchvision=0.9 with cuda version 11.1

NickNameInvalid

自定义数据集的如何设置cut_height参数

2

你好，作者： 感谢你的开源，我最近想使用自己的数据集进行模型训练，我在创建了自己的clr_resnet18_mydataset.py配置文件，我对`cut_height`参数的设置有一些疑问，这个参数的意思是将图像crop为（0，cut_height, img_w, img_h）的图像，我看TuSimple、CULane、LLAMMAS数据集都设置了`cut_height`，请这个数据的设置依据是什么？如果我不知道如何设置，是否应该把`cut_height`设置为0，即不对图像进行crop。 另外，在`clrnet/dataset`下，计算`ys`时，也用到了`cut_height`， 以及设置了一个`step`，比如CULane数据集中的`np.arange(270, 590, 8)`，其中`step=8`，请问这里step设置的依据是什么？？？ https://github.com/Turoad/CLRNet/blob/7269e9d1c1c650343b6c7febb8e764be538b1aed/clrnet/datasets/llamas.py#L146 https://github.com/Turoad/CLRNet/blob/7269e9d1c1c650343b6c7febb8e764be538b1aed/clrnet/datasets/culane.py#L97 https://github.com/Turoad/CLRNet/blob/7269e9d1c1c650343b6c7febb8e764be538b1aed/clrnet/datasets/tusimple.py#L64

cyn-liu

error ：nvcc fatal : Unknown option '-generate-dependencies-with-compile' ninja: build stopped: subcommand failed.

3

When i follow the [README.md](https://github.com/Turoad/CLRNet#readme), Try : python setup.py build develop. Then i got some erros , which i could not overcome for a few weeks . The main erro...

Q455

The size of the weights I trained myself is different from the official one.

1

Hello author, thank you for your outstanding contribution. When I trained the Tusimple dataset using "python configs/clrnet/clr_resnet18_tusimple.py --gpus 0", the weight size of the 10th epoch I obtained was 104.6MB,...

CrazyMustard-404

How do I modify this if I only want to use CULane's shadow datasets? Can I just edit the category in culane.py?

allchiever

在测试结果中为什么没有 FPS结果呢，是需要onnx还是tensorrt进行推理才能得出嘛？ 感谢您的及时回复！！！

anhuidinglieqiaodaima

你好请问我的是什么问题？Hello, what is my problem?

2

Validate: 100%|███████████████████████████████████████████████████████████████████████████████████ 166/166 [03:40

ZhuFfengJiang

请问可以区分实线和虚线吗？

7

Shicc

What is height and width of resized data?

1

Hi, i wonder height and width of resized data. I read the paper, CLRNet. It's said that 'All input images are resized to 320 x 800'. As i know, resized...

jinhoyoho