Method Details


Details for method 'EKENet'

 

Method overview

name EKENet
challenge pixel-level semantic labeling
details
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data
runtime 0.0229 s
Single TITAN X
subsampling no
submission date October, 2019
previous submissions

 

Average results

Metric Value
IoU Classes 74.2801
iIoU Classes 44.1444
IoU Categories 87.1856
iIoU Categories 69.928

 

Class results

Class IoU iIoU
road 98.2439 -
sidewalk 83.4838 -
building 91.11 -
wall 59.4676 -
fence 53.4599 -
pole 55.2612 -
traffic light 61.103 -
traffic sign 66.8774 -
vegetation 92.1978 -
terrain 70.3362 -
sky 94.7104 -
person 78.9024 56.0463
rider 59.7833 31.8356
car 93.6659 85.3192
truck 64.3431 24.17
bus 85.5251 41.1639
train 81.0046 37.8257
motorcycle 55.5685 30.1394
bicycle 66.2775 46.6554

 

Category results

Category IoU iIoU
flat 98.3536 -
nature 91.8491 -
object 62.0336 -
sky 94.7104 -
construction 91.2427 -
human 79.0621 57.2095
vehicle 93.0479 82.6465

 

Links

Download results as .csv file

Benchmark page