Method Details


Details for method 'BshapeNet+ [fine-only]'

 

Method overview

name BshapeNet+ [fine-only]
challenge instance-level semantic labeling
details BshapeNet+, ResNet-50-FPN as base model, Cityscapes [fine-only]
publication BshapeNet: Object Detection and Instance Segmentation with Bounding Shape Masks
Ba Rom Kang, Ha Young Kim
https://arxiv.org/abs/1810.10327
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date March, 2019
previous submissions

 

Average results

Metric Value
AP 27.3081
AP50% 50.4007
AP100m 40.5018
AP50m 43.1333

 

Class results

Class AP AP50% AP100m AP50m
person 29.7282 57.8343 47.8525 48.2373
rider 23.3599 57.1519 36.8103 37.2074
car 46.7393 68.6883 67.7646 70.5329
truck 26.0584 37.2001 36.692 42.3292
bus 33.3201 47.9616 49.4289 56.7411
train 24.8447 48.5825 36.2452 39.5269
motorcycle 20.3137 46.4821 27.3275 28.7563
bicycle 14.1006 39.3051 21.8931 21.735

 

Links

Download results as .csv file

Benchmark page