Method Details


Details for method 'FoveaNet'

 

Method overview

name FoveaNet
challenge pixel-level semantic labeling
details fcrn_v12_merge_v4, trained with 2975 images only.
publication Anonymous
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date August, 2016
previous submissions

 

Average results

Metric Value
IoU Classes 71.818
iIoU Classes 49.6566
IoU Categories 88.545
iIoU Categories 75.4457

 

Class results

Class IoU iIoU
road 98.1439 -
sidewalk 82.9198 -
building 91.0417 -
wall 43.0234 -
fence 49.2712 -
pole 60.2753 -
traffic light 66.9445 -
traffic sign 73.1135 -
vegetation 92.3801 -
terrain 69.6189 -
sky 94.7889 -
person 81.4404 64.1011
rider 57.1543 41.3786
car 94.2214 86.8627
truck 53.7083 29.9032
bus 64.7645 40.5397
train 60.5825 37.7409
motorcycle 59.8292 39.5993
bicycle 71.3193 57.1274

 

Category results

Category IoU iIoU
flat 98.4549 -
nature 92.1149 -
object 67.1131 -
sky 94.7889 -
construction 91.4207 -
human 83.0309 66.049
vehicle 92.8916 84.8425

 

Links

Download results as .csv file

Benchmark page