Method Details


Details for method 'Le_Selfdriving_VGG'

 

Method overview

name Le_Selfdriving_VGG
challenge pixel-level semantic labeling
details
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date September, 2016
previous submissions

 

Average results

Metric Value
IoU Classes 65.8779
iIoU Classes 35.5791
IoU Categories 84.3996
iIoU Categories 64.2706

 

Class results

Class IoU iIoU
road 97.4573 -
sidewalk 78.681 -
building 88.8892 -
wall 42.6756 -
fence 44.2393 -
pole 46.3882 -
traffic light 53.3759 -
traffic sign 61.062 -
vegetation 90.2467 -
terrain 68.5687 -
sky 93.3991 -
person 74.0937 48.5267
rider 48.522 25.3225
car 91.8972 80.9758
truck 44.9053 16.479
bus 62.4234 26.7691
train 52.342 21.6277
motorcycle 51.2128 25.0398
bicycle 61.2998 39.892

 

Category results

Category IoU iIoU
flat 98.0375 -
nature 89.8707 -
object 54.6757 -
sky 93.3991 -
construction 88.8746 -
human 75.1987 50.0327
vehicle 90.7408 78.5085

 

Links

Download results as .csv file

Benchmark page