Method Details


Details for method 'XPARSE'

 

Method overview

name XPARSE
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 November, 2016
previous submissions

 

Average results

Metric Value
IoU Classes 73.4498
iIoU Classes 49.1703
IoU Categories 88.7082
iIoU Categories 74.1772

 

Class results

Class IoU iIoU
road 98.2855 -
sidewalk 83.8522 -
building 91.587 -
wall 47.6169 -
fence 53.3946 -
pole 59.5271 -
traffic light 66.8224 -
traffic sign 72.5255 -
vegetation 92.6664 -
terrain 70.8773 -
sky 95.1955 -
person 82.4312 61.6913
rider 63.5453 40.5034
car 94.7141 87.5472
truck 57.4208 31.2733
bus 68.8313 42.7463
train 62.2178 35.8595
motorcycle 62.5585 38.3417
bicycle 71.4772 55.3999

 

Category results

Category IoU iIoU
flat 98.5257 -
nature 92.4005 -
object 66.5412 -
sky 95.1955 -
construction 91.8615 -
human 82.7698 62.954
vehicle 93.6631 85.4005

 

Links

Download results as .csv file

Benchmark page