UPSNet: A Unified Panoptic Segmentation Network

CVPR 2019 Yuwen XiongRenjie LiaoHengshuang ZhaoRui HuMin BaiErsin YumerRaquel Urtasun

In this paper, we propose a unified panoptic segmentation network (UPSNet) for tackling the newly proposed panoptic segmentation task. On top of a single backbone residual network, we first design a deformable convolution based semantic segmentation head and a Mask R-CNN style instance segmentation head which solve these two subtasks simultaneously... (read more)

PDF Abstract

Evaluation results from the paper


#2 best model for Panoptic Segmentation on Cityscapes val (using extra training data)

     Get a GitHub badge
Task Dataset Model Metric name Metric value Global rank Uses extra
training data
Compare
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101, multiscale) PQ 61.8 # 2
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101, multiscale) PQst 64.8 # 2
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101, multiscale) PQth 57.6 # 2
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101, multiscale) mIoU 79.2 # 1
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101, multiscale) AP 39.0 # 1
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101) PQ 60.5 # 5
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101) PQst 63.0 # 5
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101) PQth 57.0 # 4
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101) mIoU 77.8 # 3
Panoptic Segmentation Cityscapes val UPSNet (ResNet-101) AP 37.8 # 2
Panoptic Segmentation Cityscapes val UPSNet (ResNet-50) PQ 59.3 # 7
Panoptic Segmentation Cityscapes val UPSNet (ResNet-50) PQst 62.7 # 7
Panoptic Segmentation Cityscapes val UPSNet (ResNet-50) PQth 54.6 # 8
Panoptic Segmentation Cityscapes val UPSNet (ResNet-50) mIoU 75.2 # 7
Panoptic Segmentation Cityscapes val UPSNet (ResNet-50) AP 33.3 # 7