Method Details


Details for method 'MSNET'

 

Method overview

name MSNET
challenge pixel-level semantic labeling
details multiPath fusion
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data
runtime n/a
subsampling no
submission date October, 2017
previous submissions 1

 

Average results

Metric Value
IoU Classes 70.0838
iIoU Classes 48.3529
IoU Categories 87.4865
iIoU Categories 74.5429

 

Class results

Class IoU iIoU
road 97.7364 -
sidewalk 81.5168 -
building 90.1786 -
wall 44.0096 -
fence 50.9727 -
pole 56.9768 -
traffic light 61.8761 -
traffic sign 68.1048 -
vegetation 91.9945 -
terrain 70.0047 -
sky 93.9497 -
person 80.7511 62.4414
rider 61.1605 41.985
car 93.3323 86.2948
truck 50.6063 29.4816
bus 59.4711 41.6636
train 48.7997 30.2913
motorcycle 61.0092 38.5292
bicycle 69.142 56.1367

 

Category results

Category IoU iIoU
flat 98.224 -
nature 91.6511 -
object 63.9432 -
sky 93.9497 -
construction 90.6498 -
human 81.3655 64.2296
vehicle 92.6222 84.8563

 

Links

Download results as .csv file

Benchmark page