Method Details

Details for method 'Pixelwise Instance Segmentation with a Dynamically Instantiated Network'


Method overview

name Pixelwise Instance Segmentation with a Dynamically Instantiated Network
challenge instance-level semantic labeling
details We propose an Instance Segmentation system that produces a segmentation map where each pixel is assigned an object class and instance identity label (this has recently been termed "Panoptic Segmentation"). Our method is based on an initial semantic segmentation module which feeds into an instance subnetwork. This subnetwork uses the initial category-level segmentation, along with cues from the output of an object detector, within an end-to-end CRF to predict instances. This part of our model is dynamically instantiated to produce a variable number of instances per image. Our end-to-end approach requires no post-processing and considers the image holistically, instead of processing independent proposals. As a result, it reasons about occlusions (unlike some related work, a single pixel cannot belong to multiple instances).
publication Pixelwise Instance Segmentation with a Dynamically Instantiated Network
Anurag Arnab and Philip H. S. Torr
Computer Vision and Pattern Recognition (CVPR) 2017
project page / code
used Cityscapes data fine annotations, coarse annotations
used external data ImageNet
runtime n/a
subsampling no
submission date January, 2018
previous submissions 1


Average results

Metric Value
AP 23.4193
AP50% 45.1704
AP100m 36.8084
AP50m 40.9342


Class results

Class AP AP50% AP100m AP50m
person 20.9534 46.7858 38.4286 39.1837
rider 18.4448 48.1622 30.4267 31.1795
car 31.7009 55.783 47.5935 49.6689
truck 22.8151 33.4445 35.6365 43.2662
bus 31.0998 45.5297 49.8055 61.622
train 31.0455 53.7259 44.809 53.4576
motorcycle 19.6292 44.919 27.5078 28.6756
bicycle 11.6658 33.0129 20.2592 20.4198



Download results as .csv file

Benchmark page