Method Details


Details for method 'W3_FAFM'

 

Method overview

name W3_FAFM
challenge pixel-level semantic labeling
details Team: BOSCH-XC-DX-WAVE3
publication Junyan Yang, Qian Xu, Lei La
project page / code
used Cityscapes data fine annotations
used external data
runtime 0.029309 s
GPU: 3090
subsampling no
submission date April, 2023
previous submissions

 

Average results

Metric Value
IoU Classes 80.5241
iIoU Classes 59.2508
IoU Categories 91.0864
iIoU Categories 80.3281

 

Class results

Class IoU iIoU
road 98.6663 -
sidewalk 86.5068 -
building 93.3638 -
wall 59.3704 -
fence 59.4485 -
pole 67.439 -
traffic light 76.8795 -
traffic sign 79.2345 -
vegetation 93.6048 -
terrain 72.1496 -
sky 95.5 -
person 86.8297 71.0631
rider 73.5545 52.8746
car 95.8717 90.2941
truck 70.3161 40.1071
bus 88.0686 54.0276
train 86.1465 54.3329
motorcycle 69.7378 46.2849
bicycle 77.2697 65.0223

 

Category results

Category IoU iIoU
flat 98.6687 -
nature 93.294 -
object 73.8786 -
sky 95.5 -
construction 93.6894 -
human 87.0207 72.092
vehicle 95.5536 88.5642

 

Links

Download results as .csv file

Benchmark page