Some questiones on tensorflow version on CULane

[hustcc19860606]

I try to test tensorflow version on CULane. But, i can not reproduct the result by command:

' root@af654694758d:/home/cc/SCNN-Tensorflow/tools/lane_evaluation# ./evaluate -a /home/cc/SCNN-Tensorflow/data/CULane/ -d /home/cc/SCNN-Tensorflow/data/CULane/result-matlab/scnn/vgg_SCNN_DULR_w9/data/CULane/ -i /home/cc/SCNN-Tensorflow/data/CULane/ -l /home/cc/SCNN-Tensorflow/data/CULane/test.txt -w 30 -t 0.5 -c 1640 -r 590 -f 1 -o ./output/test.txt '

The result is : list_im_file: /home/cc/SCNN-Tensorflow/data/CULane/test.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590 Evaluating the results... tp: 0 fp: 106172 fn: 104886 finished process file precision: 0 recall: 0 Fmeasure: -nan

My questiones has two:

1. there is no need to modify code or parameters to follow SCNN to get curve lines from probability maps as well as calculate precision, recall and F1-measure?

2.In your README.md, "Note that you need to exchange the order of VGG-MEAN in test_lanenet.py and change the order of input images from RGB to BGR since the pre-trained model uses opencv to read images." In test_lanenet.py, VGG_MEAN = [103.939, 116.779, 123.68], Do I need to change it to VGG_MEAN = [123.68, 116.779, 103.939]?

@XingangPan

[XingangPan]

@hustcc19860606 I am not very familiar with the tensorflow version, maybe you can open an issue under https://github.com/cardwing/Codes-for-Lane-Detection? But I think you still need to get curve lines from probability maps so as to calculate accuracies.

[hustcc19860606]

@hustcc19860606 I am not very familiar with the tensorflow version, maybe you can open an issue under https://github.com/cardwing/Codes-for-Lane-Detection? But I think you still need to get curve lines from probability maps so as to calculate accuracies.

Thanks， Is the curve lines stored in txt, like‘driver_37_30frame\05181432_0203.MP4\00240.lines.txt’？ 00240.lines.txt

[XingangPan]

Yes.