Method Details


Details for method 'Seg_UJS'

 

Method overview

name Seg_UJS
challenge pixel-level semantic labeling
details Previously listed as "DLnet".
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date December, 2019
previous submissions

 

Average results

Metric Value
IoU Classes 83.3814
iIoU Classes 64.327
IoU Categories 92.1523
iIoU Categories 81.9829

 

Class results

Class IoU iIoU
road 98.7844 -
sidewalk 87.7909 -
building 94.1563 -
wall 64.055 -
fence 65.0003 -
pole 72.0171 -
traffic light 78.8164 -
traffic sign 82.6549 -
vegetation 94.1581 -
terrain 73.9816 -
sky 96.1169 -
person 88.1084 72.7696
rider 75.3041 55.1467
car 96.4229 92.2119
truck 78.7772 50.1164
bus 94.0106 60.9446
train 91.5358 62.9854
motorcycle 73.6689 53.7476
bicycle 78.8863 66.6939

 

Category results

Category IoU iIoU
flat 98.8017 -
nature 93.9039 -
object 77.689 -
sky 96.1169 -
construction 94.3474 -
human 88.1021 73.4698
vehicle 96.1053 90.4959

 

Links

Download results as .csv file

Benchmark page