caffemodel for VPGNet

[UcanSee]

I have trained a caffemodel from your tutorial, but the test result is bad. I think it is because of little dataset. Could you please tell me where I can download the lane and road marking benchmark which consists of about 20,000 images or give me the caffemodel described in your paper? Thank you very much!

[SeokjuLee]

Hi, the dataset and model are preserved by Samsung Electronics. Please contact Tae-Hee Lee ([email protected]), Hyun Seok Hong ([email protected]), and Seung-Hoon Han ([email protected]). Thanks!

[doem97]

@SeokjuLee Hi Dr. SeojkuLee, can you specify what is the data format with VP information in it?

e.g. for no-VP, file list data format is like VPGdata/000001.mat 3 536 184 544 192 15 544 184 552 192 15 552 184 560 192 15 ... which is "filename, label_grid_num, x1a, y1a, x1b, y1b, label_1, x2a, y2a, x2b, label_2, ... xna, yna, xnb, ynb, label_n" You combine road_marking and lane label information into single label_n digit, from 1-8 is lane; from 9-17 is road marking.

Can you show me if I want to extract the VP information myself, how can I add it into this data format?

[doem97]

Or could you please show me any instance? only one sample image's label file is good.

[sandeepnmenon]

@doem97 Were you able to obtain the instance and the trained model?