Method Details


Details for method 'TuSimple'

 

Method overview

name TuSimple
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 October, 2016
previous submissions 1, 2

 

Average results

Metric Value
IoU Classes 76.761
iIoU Classes 52.18
IoU Categories 90.0256
iIoU Categories 75.9645

 

Class results

Class IoU iIoU
road 98.4491 -
sidewalk 85.0736 -
building 92.6191 -
wall 55.6848 -
fence 54.3812 -
pole 64.3109 -
traffic light 72.5836 -
traffic sign 77.1454 -
vegetation 93.2669 -
terrain 71.4299 -
sky 95.2845 -
person 84.6065 64.3084
rider 66.5019 43.1995
car 95.5014 88.7443
truck 69.8336 36.5621
bus 75.4222 46.9302
train 67.9459 40.7505
motorcycle 65.0962 40.4574
bicycle 73.3223 56.4878

 

Category results

Category IoU iIoU
flat 98.5974 -
nature 92.9061 -
object 71.1913 -
sky 95.2845 -
construction 92.8025 -
human 84.8891 65.5007
vehicle 94.5082 86.4284

 

Links

Download results as .csv file

Benchmark page