Method Details


Details for method 'DAHUA-ARI'

 

Method overview

name DAHUA-ARI
challenge pixel-level semantic labeling
details mult-scale and refine
publication Anonymous
project page / code
used Cityscapes data fine annotations, coarse annotations
used external data ImageNet, Mapillary
runtime n/a
subsampling no
submission date October, 2020
previous submissions

 

Average results

Metric Value
IoU Classes 85.6144
iIoU Classes 70.3868
IoU Categories 93.186
iIoU Categories 85.5486

 

Class results

Class IoU iIoU
road 99.0043 -
sidewalk 89.534 -
building 94.9543 -
wall 71.7526 -
fence 69.1317 -
pole 75.8487 -
traffic light 82.166 -
traffic sign 85.2368 -
vegetation 94.5305 -
terrain 75.418 -
sky 96.3323 -
person 90.0424 78.6076
rider 79.4289 63.2473
car 96.9362 92.7327
truck 81.7283 57.7306
bus 95.0209 71.0687
train 91.1699 65.3185
motorcycle 77.0094 61.8406
bicycle 81.4294 72.5487

 

Category results

Category IoU iIoU
flat 98.9281 -
nature 94.3507 -
object 80.8499 -
sky 96.3323 -
construction 95.0124 -
human 90.1126 79.4491
vehicle 96.7161 91.6481

 

Links

Download results as .csv file

Benchmark page