-
I have been reading the source code of etfnet-culane-pytorch for two days, but I cannot find the code of self-attention distillation in which module, could you please tell me?
-
For Curve category, the F1 measure of ERFNet should be 66.3 instead of 71.6
https://github.com/cardwing/Codes-for-Lane-Detection
I ran the evaluation code on your trained model, and obtained exact…
-
你好，请教下你自己训练的模型在CULane上最后损失大概在多少；我把初始学习率调整到5e-4 ，一个epoch之后基本已经不下降了，大概在0.14，我之前自己用tensorflow训练过TuSimple的训练集，用分割的交叉熵损失，最后也在0.05~0.1；但我不确定在CULane上损失可以降到多少？这两个训练集我用的都是自己搭的网络。
-
**你好，对于SCNN中的CULane数据集，目前我有三个问题，向您请教。**
**问题1：**从test后得到的概率图中，我发现所有的lines.txt中，都存储的是“1”，请问你有了解过这个问题么？
**问题2：**在最后的评估中，我发现对Crossroad情景下车道线评估错误，我想知道为什么会出现这种情况？因为我是遵循您的SCNN训练模型进行的测试，错误评估结果如下：
model/CU…
-
Hi XingangPan,
I am wondering in these .line.txt why the coordinates, x is a number contain decimal point but y is integer? Does this design have special meaning?
Thanks
-
when training the erfnet on CuLane, I always encounter the tensor dimension mismatch problem. I add the ceil_mode = True in the maxpool layer and the problem becomes as follows:
RuntimeError: inva…
-
https://github.com/cardwing/Codes-for-Lane-Detection/blob/8a04dd40b0fd53fe5348529af060871d401ee987/ERFNet-CULane-PyTorch/test_erfnet.py#L13
running `sh ./test_erfnet.sh` leaves me with mentioned er…
-
Sorry to ask you that how to set up CULane datasets for training ?
I have already read all of issues concerning with CULane, but didn't find the suggestions how to set up CULane datasets. Hope you…
-
CULane数据集中，laneseg_label_w16文件中所有图片都是黑色的，如下所示：
之后，我又检查了下train_gt.txt，里面的路径，如下所示：
![图片](ht…
-
@cardwing @yujincheng08 ,On CULane for SCNN，我想使用2个gpu进行训练，因此我按照你给出的方法，如下所示：
Multi-GPU training has been supported. Just change BATCH_SIZE and GPU_NUM in global_config.py, and then use CUDA_VISIBLE_DE…