Method Details


Details for method 'RMNet'

 

Method overview

name RMNet
challenge pixel-level semantic labeling
details A fast and light net for semantic segmentation.
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data
runtime 0.014 s
GPU
subsampling no
submission date April, 2018
previous submissions

 

Average results

Metric Value
IoU Classes 64.4618
iIoU Classes 37.3235
IoU Categories 84.5872
iIoU Categories 67.7412

 

Class results

Class IoU iIoU
road 97.5079 -
sidewalk 79.3478 -
building 88.5011 -
wall 37.0308 -
fence 40.1823 -
pole 52.6294 -
traffic light 55.6742 -
traffic sign 58.2178 -
vegetation 90.5027 -
terrain 67.2232 -
sky 93.3234 -
person 72.8077 51.4873
rider 51.2015 28.9839
car 91.4004 83.9024
truck 42.5287 19.4248
bus 53.6568 28.7585
train 52.6539 24.1769
motorcycle 41.8686 21.6862
bicycle 58.5152 40.1679

 

Category results

Category IoU iIoU
flat 98.0085 -
nature 90.2129 -
object 58.1342 -
sky 93.3234 -
construction 88.4916 -
human 73.6243 53.534
vehicle 90.3152 81.9485

 

Links

Download results as .csv file

Benchmark page