Method Details

Details for method 'Panoptic-DeepLab [Cityscapes-fine]'


Method overview

name Panoptic-DeepLab [Cityscapes-fine]
challenge instance-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. This submission exploits only Cityscapes fine annotations. This entry fixes a minor inference bug (i.e., same trained model) for instance segmentation, compared to the previous submission.
publication Panoptic-DeepLab
Bowen Cheng, Maxwell D. Collins, Yukun Zhu, Ting Liu, Thomas S. Huang, Hartwig Adam, Liang-Chieh Chen
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date October, 2019
previous submissions 1


Average results

Metric Value
AP 34.6462
AP50% 57.2669
AP100m 50.4717
AP50m 53.133


Class results

Class AP AP50% AP100m AP50m
person 34.3287 63.883 54.1593 54.2908
rider 28.8696 61.1347 43.8113 44.6366
car 55.0604 77.8674 76.3908 78.8617
truck 32.8259 42.0154 47.7383 52.7469
bus 41.4894 53.3573 63.3842 70.2216
train 36.6418 56.0755 49.2922 54.3213
motorcycle 26.3188 53.2601 35.3858 36.3494
bicycle 21.6352 50.5418 33.612 33.6361



Download results as .csv file

Benchmark page