Method Details


Details for method 'PSPNet'

 

Method overview

name PSPNet
challenge pixel-level semantic labeling
details Using training set only (2975 images) without adding validation set or coarse set, no post-processsing like CRF.
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date June, 2016
previous submissions 1

 

Average results

Metric Value
IoU Classes 74.6331
iIoU Classes 51.0583
IoU Categories 89.1166
iIoU Categories 74.9849

 

Class results

Class IoU iIoU
road 98.3871 -
sidewalk 84.5362 -
building 91.5428 -
wall 40.6248 -
fence 53.4612 -
pole 59.8502 -
traffic light 72.1072 -
traffic sign 75.9065 -
vegetation 92.6021 -
terrain 71.6697 -
sky 94.3782 -
person 84.5027 62.9113
rider 67.3873 42.701
car 95.2976 88.5181
truck 61.9827 32.5195
bus 70.6117 44.9804
train 61.4964 38.881
motorcycle 66.6166 40.5815
bicycle 75.0672 57.3737

 

Category results

Category IoU iIoU
flat 98.5932 -
nature 92.2682 -
object 67.5248 -
sky 94.3782 -
construction 91.9798 -
human 85.1343 63.9551
vehicle 93.9379 86.0147

 

Links

Download results as .csv file

Benchmark page