Method Details


Details for method 'Panoptic-DeepLab [Mapillary Vistas]'

 

Method overview

name Panoptic-DeepLab [Mapillary Vistas]
challenge pixel-level semantic labeling
details Our proposed bottom-up Panoptic-DeepLab is conceptually simple yet delivers state-of-the-art results. The Panoptic-DeepLab adopts dual-ASPP and dual-decoder modules, specific to semantic segmentation and instance segmentation respectively. The semantic segmentation prediction follows the typical design of any semantic segmentation model (e.g., DeepLab), while the instance segmentation prediction involves a simple instance center regression, where the model learns to predict instance centers as well as the offset from each pixel to its corresponding center.
publication Panoptic-DeepLab
Bowen Cheng, Maxwell D. Collins, Yukun Zhu, Ting Liu, Thomas S. Huang, Hartwig Adam, Liang-Chieh Chen
https://arxiv.org/abs/1910.04751
project page / code
used Cityscapes data fine annotations
used external data ImageNet, Mapillary Vistas Research Edition.
runtime n/a
subsampling no
submission date September, 2019
previous submissions

 

Average results

Metric Value
IoU Classes 84.1675
iIoU Classes 66.8067
IoU Categories 92.6892
iIoU Categories 82.0257

 

Class results

Class IoU iIoU
road 98.8018 -
sidewalk 88.0819 -
building 94.4917 -
wall 68.1057 -
fence 68.0888 -
pole 74.532 -
traffic light 80.5057 -
traffic sign 83.5256 -
vegetation 94.1752 -
terrain 74.4237 -
sky 96.1427 -
person 89.159 74.9374
rider 77.0801 58.6193
car 96.5012 89.3491
truck 78.9006 52.6863
bus 91.8356 66.5983
train 89.1442 66.2167
motorcycle 76.4357 59.9849
bicycle 79.2506 66.0615

 

Category results

Category IoU iIoU
flat 98.8364 -
nature 93.9891 -
object 79.56 -
sky 96.1427 -
construction 94.6683 -
human 89.3171 75.9263
vehicle 96.3106 88.1251

 

Links

Download results as .csv file

Benchmark page