Method Details


Details for method 'NAVINFO_DLR'

 

Method overview

name NAVINFO_DLR
challenge pixel-level semantic labeling
details weighted aspp+ohem+hard region refine
publication pengfei zhang
project page / code
used Cityscapes data fine annotations, coarse annotations
used external data ImageNet
runtime n/a
subsampling no
submission date November, 2019
previous submissions 1

 

Average results

Metric Value
IoU Classes 83.8074
iIoU Classes 65.615
IoU Categories 92.4389
iIoU Categories 83.7423

 

Class results

Class IoU iIoU
road 98.8247 -
sidewalk 87.8713 -
building 94.1645 -
wall 65.1579 -
fence 63.5298 -
pole 73.4076 -
traffic light 80.1515 -
traffic sign 83.0366 -
vegetation 94.1369 -
terrain 73.48 -
sky 95.8599 -
person 88.9662 76.5365
rider 76.8016 57.8462
car 96.6525 91.3601
truck 82.0739 53.2327
bus 93.7105 61.8313
train 90.4968 61.2277
motorcycle 74.068 54.5703
bicycle 79.9509 68.3149

 

Category results

Category IoU iIoU
flat 98.816 -
nature 93.8774 -
object 78.711 -
sky 95.8599 -
construction 94.4048 -
human 89.0614 77.4467
vehicle 96.3417 90.038

 

Links

Download results as .csv file

Benchmark page