Can't the performance of the algorithm be reproduced on curvelanes?

[Jinming-Su]

I only got poor performance with F1-measure 82.90% by using the official code provided here (curvelanes small), which is far worse than the checkpoint provided with F1-measure 85.09%. Please help me answer why.

[ztjsw]

@Jinming-Su I also get the similar F1 and cannot reproduce the result.

{'f1_measure': 0.8287583304495372, 'precision': 0.8762067744600496, 'recall': 0.7861847678752966, 'pr_num': 84730, 'gt_num': 94432

[lovelydjj]

@ztjsw @Jinming-Su When I reproduce in curvelanes, I first changed the directory, and then entered the command
”python tools/condlanenet/curvelanes/test_curvelanes.py configs/condlanenet/curvelanes/curvelanes_small_test.py ./weight_path/curvelanes/curvelanes_small.pth --show“
then I encountered a problem

. I would like to ask you how to solve it. If you can answer, I will be very grateful