Method Details


Details for method 'SAC-multiple'

 

Method overview

name SAC-multiple
challenge pixel-level semantic labeling
details
publication Scale-adaptive Convolutions for Scene Parsing
Rui Zhang, Sheng Tang, Yongdong Zhang, Jintao Li, and Shuicheng Yan
International Conference on Computer Vision (ICCV) 2017
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date March, 2017
previous submissions

 

Average results

Metric Value
IoU Classes 78.1268
iIoU Classes 55.2327
IoU Categories 90.5768
iIoU Categories 78.3172

 

Class results

Class IoU iIoU
road 98.6809 -
sidewalk 86.5434 -
building 93.1452 -
wall 56.2978 -
fence 59.5215 -
pole 65.0954 -
traffic light 72.9547 -
traffic sign 78.2459 -
vegetation 93.5051 -
terrain 72.5836 -
sky 95.5684 -
person 85.8973 67.0492
rider 70.8167 48.3739
car 95.9403 90.2815
truck 71.154 39.3918
bus 78.6001 50.5651
train 66.1786 41.9886
motorcycle 67.6538 43.5414
bicycle 76.0257 60.6702

 

Category results

Category IoU iIoU
flat 98.6883 -
nature 93.1412 -
object 71.7713 -
sky 95.5684 -
construction 93.4471 -
human 86.1227 68.4076
vehicle 95.2985 88.2269

 

Links

Download results as .csv file

Benchmark page