Method Details


Details for method 'LDN2_ROB'

 

Method overview

name LDN2_ROB
challenge pixel-level semantic labeling
details Ladder DenseNet: https://ivankreso.github.io/publication/ladder-densenet/
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data ImageNet, ROB challenge data
runtime 1 s
subsampling no
submission date June, 2018
previous submissions

 

Average results

Metric Value
IoU Classes 77.8916
iIoU Classes 52.256
IoU Categories 90.0974
iIoU Categories 77.0835

 

Class results

Class IoU iIoU
road 98.4971 -
sidewalk 85.1266 -
building 92.7674 -
wall 55.6081 -
fence 56.7435 -
pole 63.7053 -
traffic light 70.9104 -
traffic sign 76.0778 -
vegetation 93.302 -
terrain 71.3351 -
sky 95.615 -
person 85.014 65.0724
rider 68.8104 42.8952
car 95.4986 90.098
truck 74.8821 34.8502
bus 80.9565 48.2115
train 74.7423 39.7655
motorcycle 66.5217 38.9217
bicycle 73.8262 58.2336

 

Category results

Category IoU iIoU
flat 98.6458 -
nature 92.9567 -
object 70.1892 -
sky 95.615 -
construction 93.0179 -
human 85.2439 66.0808
vehicle 95.0132 88.0862

 

Links

Download results as .csv file

Benchmark page