Method Details


Details for method 'HRNet + LKPP + EA loss'

 

Method overview

name HRNet + LKPP + EA loss
challenge pixel-level semantic labeling
details
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data
runtime n/a
subsampling no
submission date August, 2020
previous submissions

 

Average results

Metric Value
IoU Classes 82.5874
iIoU Classes 61.7971
IoU Categories 91.3557
iIoU Categories 78.5906

 

Class results

Class IoU iIoU
road 98.7935 -
sidewalk 87.4973 -
building 93.8153 -
wall 67.2949 -
fence 63.7468 -
pole 68.4762 -
traffic light 75.8824 -
traffic sign 80.45 -
vegetation 93.7471 -
terrain 71.9212 -
sky 95.6171 -
person 87.2131 66.6098
rider 73.617 51.0147
car 96.3651 91.3404
truck 82.9663 47.045
bus 92.7232 61.6416
train 88.508 60.6044
motorcycle 72.8224 51.4369
bicycle 77.7042 64.6836

 

Category results

Category IoU iIoU
flat 98.7636 -
nature 93.515 -
object 74.5814 -
sky 95.6171 -
construction 93.9794 -
human 87.1204 67.6096
vehicle 95.9129 89.5716

 

Links

Download results as .csv file

Benchmark page