Method Details


Details for method 'EFBNET'

 

Method overview

name EFBNET
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 January, 2018
previous submissions

 

Average results

Metric Value
IoU Classes 81.7849
iIoU Classes 59.9386
IoU Categories 90.7189
iIoU Categories 78.7883

 

Class results

Class IoU iIoU
road 98.6305 -
sidewalk 86.6309 -
building 93.3574 -
wall 60.6249 -
fence 63.3632 -
pole 65.3928 -
traffic light 75.3858 -
traffic sign 79.5552 -
vegetation 93.5063 -
terrain 72.141 -
sky 95.153 -
person 86.5126 68.7194
rider 72.9142 49.7729
car 95.996 90.1948
truck 81.3932 43.0285
bus 92.727 55.4987
train 89.7855 60.9751
motorcycle 73.62 49.0145
bicycle 77.2226 62.3053

 

Category results

Category IoU iIoU
flat 98.632 -
nature 93.192 -
object 72.5034 -
sky 95.153 -
construction 93.6295 -
human 86.3978 69.3642
vehicle 95.5249 88.2124

 

Links

Download results as .csv file

Benchmark page