Method Details


Details for method 'FANet'

 

Method overview

name FANet
challenge pixel-level semantic labeling
details Feature Aggregation Network for Semantic Segmentation
publication FANet: Feature Aggregation Network for Semantic Segmentation
Tanmay Singha, Duc-Son Pham, and Aneesh Krishna
https://ieeexplore.ieee.org/abstract/document/9363370
project page / code
used Cityscapes data fine annotations
used external data
runtime n/a
subsampling no
submission date July, 2020
previous submissions

 

Average results

Metric Value
IoU Classes 64.0814
iIoU Classes 33.2002
IoU Categories 83.0867
iIoU Categories 61.0919

 

Class results

Class IoU iIoU
road 96.7403 -
sidewalk 75.3459 -
building 88.1509 -
wall 35.3957 -
fence 37.8316 -
pole 45.6847 -
traffic light 51.3414 -
traffic sign 57.4483 -
vegetation 90.4131 -
terrain 64.2628 -
sky 92.9834 -
person 71.7687 42.1324
rider 50.3763 22.2712
car 91.6062 81.7667
truck 48.9252 14.2354
bus 62.0055 26.3841
train 51.9677 20.7658
motorcycle 46.2943 20.0892
bicycle 59.0039 37.9571

 

Category results

Category IoU iIoU
flat 96.1909 -
nature 89.816 -
object 52.1872 -
sky 92.9834 -
construction 88.1205 -
human 72.2217 43.362
vehicle 90.087 78.8218

 

Links

Download results as .csv file

Benchmark page