Method Details


Details for method 'GPSNet'

 

Method overview

name GPSNet
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, 2019
previous submissions

 

Average results

Metric Value
IoU Classes 82.114
iIoU Classes 62.4327
IoU Categories 91.9583
iIoU Categories 81.599

 

Class results

Class IoU iIoU
road 98.8535 -
sidewalk 88.1351 -
building 93.8746 -
wall 61.525 -
fence 63.002 -
pole 71.5283 -
traffic light 79.0389 -
traffic sign 81.3826 -
vegetation 94.0365 -
terrain 73.1906 -
sky 95.7522 -
person 88.0798 72.9103
rider 74.9115 55.069
car 96.5289 91.2695
truck 75.5309 44.7095
bus 91.211 58.2964
train 84.1283 57.9387
motorcycle 70.6707 52.0716
bicycle 78.7851 67.1968

 

Category results

Category IoU iIoU
flat 98.8002 -
nature 93.7779 -
object 77.0317 -
sky 95.7522 -
construction 94.1989 -
human 88.0784 73.7268
vehicle 96.0688 89.4711

 

Links

Download results as .csv file

Benchmark page