Method Details


Details for method 'PIDNet'

 

Method overview

name PIDNet
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 November, 2024
previous submissions

 

Average results

Metric Value
IoU Classes 76.4363
iIoU Classes 52.3821
IoU Categories 90.1176
iIoU Categories 77.5894

 

Class results

Class IoU iIoU
road 98.4599 -
sidewalk 85.2492 -
building 92.5114 -
wall 46.5715 -
fence 56.3521 -
pole 64.6526 -
traffic light 73.3806 -
traffic sign 75.343 -
vegetation 93.2204 -
terrain 71.6659 -
sky 95.4036 -
person 84.8636 66.6918
rider 68.8703 46.6013
car 95.4884 89.1654
truck 67.0027 32.0741
bus 77.0518 43.4017
train 68.9218 40.227
motorcycle 64.1165 40.8771
bicycle 73.1645 60.0183

 

Category results

Category IoU iIoU
flat 98.6106 -
nature 92.9394 -
object 70.9645 -
sky 95.4036 -
construction 92.9767 -
human 85.1149 68.0891
vehicle 94.8137 87.0898

 

Links

Download results as .csv file

Benchmark page