chrisding/seal

Code for Simultaneous Edge Alignment and Learning (SEAL)

seal's Issues

eval the cityscape something is wrong , some score get 0

Dear ,
Thanks for your greet job! I get some trouble when eval the cityscape ,result as follow
====================== Summary MF-ODS ======================

1 Road: 0.942
2 Sidewalk: 0.896
3 Building: 0.829
4 Wall: 0.000
5 Fence: 0.052
6 Pole: 0.803
7 Traffic Light: 0.000
8 Traffic Sign: 0.821
9 Vegetation: 0.785
10 Terrain: 0.000
11 Sky: 0.949
12 Person: 0.876
13 Rider: 0.000
14 Car: 0.804
15 Truck: 0.000
16 Bus: 0.000
17 Train: 0.000
18 Motorcycle: 0.000
19 Bicycle: 0.000

  Mean MF-ODS:  0.408

yours
xu

non-instance edge prediction model

hi, could you provide the link for non-instance edge prediction pretrained models for sbd and cityscapes?

The original paper is too difficult for me~

hi. this good work is difficult for me, especially the part of "Edge prior model". Could you provide a detail version of method in Chinese such as published in Zhihu. :)