cfzd
cfzd
@nithinme3 Yes, it is correct.
@nithinme3 It is exactly the same for flipping the output and flipping the row anchor, so it is correct. For the shifted problem, it might be because the row anchor...
@nithinme3 Yes, it should work right. In the `demo.py`, there are codes that exactly match your case(Tusimple & 288x800 inference). You can refer to it.
@SmallBirdMing 没有`pre_computed_dct_weights`这个函数
@easternblood dali只支持linux，很抱歉，这个是dali自己的问题，我也没有办法
@cedricnam Sorry, I missed this problem. @byungjikim You need to modify the `data_root` in the config files according to your dataset.
@PilotPhil The top1, top2, and top3 are calculated during training to monitor the training process. For evaluation, we use the metrics defined by the dataset. If you want to get...
@AssassinJay 抱歉，我没有做过这方面。
@ennheng 这个是由于没有安装CULane评估工具导致的，可以参考`INSTALL.md`中的第五个部分安装
@ennheng 这个当然是可以的，你可以随便换成别的backbone，但是相应的你可能需要调一下参才能获得在别的backbone上最好的性能