-
作者您好！我想问下您在用ERFNet来训练CULane时是否用到了什么训练技巧，我在用ERFNet训练Tusimple数据集时，训练了60k，但结果总是不是很理想，训练集和测试集得出iou之间相差很大，训练集上iou达到0.96，但测试集只有0.47，我的学习速率和weight_decay这些参数设置都与您设置的一样
-
Hello, may I ask how to generate the label for the Backprojection_Loss training? I calculated the parameters as y=ax^2+bx+c for the Tusimple dataset, but the parameters were different from the Curve_p…
-
Hi @wvangansbeke ,
Great work!
I downloaded the sample Tusimple dataset provided in the repo. It has only two folders namely images and groundtruth.
After this, I ran the code using below comma…
-
Hi, I use the models you provided and get the same results. could you provide the Res-34 models or share the hyper-parameters when training the Res-34 models ? Because I trained the Res-34 but the r…
-
Hello, firstly thank you for your great work. After I look through the code and the data set (tuSimple), I am confused that if the overfitting problem would occur ? Because i notice that you train the…
-
Thanks you for your contribution.
In your code,using 18 row anchors for culane dataset.
can i use more row anchors for culane dataset so that the structure loss can work well? such as 36 or more.
-
HI, I use the pretrained model_2695.pt to test my own dataset.but it shows like this:
 (E2E-LMD) whose core idea is coincidentally similar to yours, which was published just one month afte…
-
Hello,thanks for your great work,
when i train polyLaneNet with my dataset about 30000 images within about 50 epochs
i run the demo like this, what's the problem with it？
-
Hello, when I run the code in theBackprojection_Loss directory on the complete dataset of TuSimple, I found that there is no "ground_truth" directory in Tusimple's train set(just a 'clips' and 3 json …