Method Details


Details for method 'SwiftNetRN-18 pyramid'

 

Method overview

name SwiftNetRN-18 pyramid
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, 2018
previous submissions

 

Average results

Metric Value
IoU Classes 74.3584
iIoU Classes 48.4225
IoU Categories 89.4714
iIoU Categories 74.5897

 

Class results

Class IoU iIoU
road 98.4427 -
sidewalk 84.6598 -
building 92.0038 -
wall 48.1931 -
fence 50.8877 -
pole 62.7108 -
traffic light 69.0645 -
traffic sign 74.5324 -
vegetation 92.5965 -
terrain 69.8212 -
sky 95.2625 -
person 83.8215 62.7025
rider 66.2436 41.3172
car 95.4449 87.7201
truck 64.7649 30.3161
bus 75.3978 39.9458
train 55.9072 34.2559
motorcycle 60.665 35.4907
bicycle 72.3895 55.6314

 

Category results

Category IoU iIoU
flat 98.5652 -
nature 92.2977 -
object 69.125 -
sky 95.2625 -
construction 92.2172 -
human 84.2501 64.1924
vehicle 94.5821 84.987

 

Links

Download results as .csv file

Benchmark page