Method Details


Details for method 'Tencent AI Lab'

 

Method overview

name Tencent AI Lab
challenge pixel-level semantic labeling
details
publication Anonymous
project page / code
used Cityscapes data fine annotations, coarse annotations
used external data Mapillary
runtime n/a
subsampling no
submission date August, 2018
previous submissions

 

Average results

Metric Value
IoU Classes 82.9077
iIoU Classes 63.8654
IoU Categories 91.8096
iIoU Categories 80.3627

 

Class results

Class IoU iIoU
road 98.636 -
sidewalk 86.8722 -
building 94.1008 -
wall 63.5187 -
fence 62.9979 -
pole 70.6885 -
traffic light 77.6669 -
traffic sign 80.2006 -
vegetation 93.9884 -
terrain 73.1068 -
sky 95.865 -
person 87.8255 71.5105
rider 74.5279 55.7879
car 96.2821 90.1086
truck 82.7878 50.6189
bus 94.2807 63.1725
train 90.428 56.1296
motorcycle 73.9747 57.0577
bicycle 77.4982 66.5377

 

Category results

Category IoU iIoU
flat 98.6944 -
nature 93.613 -
object 76.9576 -
sky 95.865 -
construction 94.1435 -
human 87.5559 72.2808
vehicle 95.8377 88.4446

 

Links

Download results as .csv file

Benchmark page