Method Details


Details for method 'MSNET'

 

Method overview

name MSNET
challenge pixel-level semantic labeling
details
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data
runtime 0.2 s
gtx1070
subsampling no
submission date October, 2017
previous submissions

 

Average results

Metric Value
IoU Classes 67.3541
iIoU Classes 45.0962
IoU Categories 85.9777
iIoU Categories 71.1471

 

Class results

Class IoU iIoU
road 97.5666 -
sidewalk 80.2975 -
building 89.351 -
wall 43.7783 -
fence 48.1256 -
pole 51.5149 -
traffic light 58.099 -
traffic sign 64.1374 -
vegetation 91.5588 -
terrain 69.1299 -
sky 93.1756 -
person 78.2178 57.8841
rider 55.2865 36.2237
car 92.7463 84.3177
truck 45.1327 27.8684
bus 54.3181 39.8107
train 42.6871 27.859
motorcycle 57.6076 33.9228
bicycle 66.9967 52.8831

 

Category results

Category IoU iIoU
flat 98.1178 -
nature 91.1443 -
object 59.4088 -
sky 93.1756 -
construction 89.8041 -
human 78.6989 59.3088
vehicle 91.4941 82.9854

 

Links

Download results as .csv file

Benchmark page