Method Details


Details for method 'ERSNet-coarse'

 

Method overview

name ERSNet-coarse
challenge pixel-level semantic labeling
details
publication Anonymous
project page / code
used Cityscapes data fine annotations, coarse annotations
used external data
runtime 0.012 s
gpu titan XP
subsampling 4
submission date January, 2019
previous submissions

 

Average results

Metric Value
IoU Classes 67.6093
iIoU Classes 39.1894
IoU Categories 85.9126
iIoU Categories 67.8346

 

Class results

Class IoU iIoU
road 97.8965 -
sidewalk 81.2697 -
building 89.8202 -
wall 44.9676 -
fence 43.899 -
pole 54.3932 -
traffic light 56.5368 -
traffic sign 63.3262 -
vegetation 91.4862 -
terrain 68.0146 -
sky 94.4996 -
person 75.0987 51.5607
rider 53.6288 28.1154
car 92.9143 85.0786
truck 50.1959 20.4423
bus 66.8959 33.4515
train 55.0354 29.4829
motorcycle 44.4158 24.0425
bicycle 60.2827 41.3414

 

Category results

Category IoU iIoU
flat 98.1695 -
nature 91.0801 -
object 60.5256 -
sky 94.4996 -
construction 89.877 -
human 75.4648 53.1123
vehicle 91.7718 82.557

 

Links

Download results as .csv file

Benchmark page