Method Details


Details for method 'ddl_seg'

 

Method overview

name ddl_seg
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 December, 2020
previous submissions

 

Average results

Metric Value
IoU Classes 84.5846
iIoU Classes 69.4181
IoU Categories 92.7584
iIoU Categories 84.4475

 

Class results

Class IoU iIoU
road 98.9713 -
sidewalk 89.2372 -
building 94.7537 -
wall 69.795 -
fence 67.2038 -
pole 73.7137 -
traffic light 76.403 -
traffic sign 82.2031 -
vegetation 94.4369 -
terrain 74.8616 -
sky 96.1653 -
person 89.7118 77.1255
rider 77.417 60.6897
car 96.9069 91.9547
truck 81.402 57.7887
bus 95.1262 70.7804
train 91.8691 65.2224
motorcycle 76.0928 61.098
bicycle 80.838 70.6855

 

Category results

Category IoU iIoU
flat 98.9187 -
nature 94.2542 -
object 78.8144 -
sky 96.1653 -
construction 94.8522 -
human 89.6315 77.9478
vehicle 96.6724 90.9473

 

Links

Download results as .csv file

Benchmark page