Fast Panoptic Segmentation Network

10/09/2019
by   Daan de Geus, et al.
15

In this work, we present an end-to-end network for fast panoptic segmentation. This network, called Fast Panoptic Segmentation Network (FPSNet), does not require computationally costly instance mask predictions or merging heuristics. This is achieved by casting the panoptic task into a custom dense pixel-wise classification task, which assigns a class label or an instance id to each pixel. We evaluate FPSNet on the Cityscapes and Pascal VOC datasets, and find that FPSNet is faster than existing panoptic segmentation methods, while achieving better or similar panoptic segmentation performance. On the Cityscapes validation set, we achieve a Panoptic Quality score of 55.1 prediction times of 114 milliseconds for images with a resolution of 1024x2048 pixels. For lower resolutions of the Cityscapes dataset and for the Pascal VOC dataset, FPSNet runs at 22 and 35 frames per second, respectively.

READ FULL TEXT

page 1

page 3

page 7

page 8

research
02/07/2019

Single Network Panoptic Segmentation for Street Scene Understanding

In this work, we propose a single deep neural network for panoptic segme...
research
07/11/2021

SE-PSNet: Silhouette-based Enhancement Feature for Panoptic Segmentation Network

Recently, there has been a panoptic segmentation task combining semantic...
research
09/28/2020

Learning Category- and Instance-Aware Pixel Embedding for Fast Panoptic Segmentation

Panoptic segmentation (PS) is a complex scene understanding task that re...
research
07/07/2020

Instance Segmentation for Whole Slide Imaging: End-to-End or Detect-Then-Segment

Automatic instance segmentation of glomeruli within kidney Whole Slide I...
research
03/27/2022

Deep Hierarchical Semantic Segmentation

Humans are able to recognize structured relations in observation, allowi...
research
05/09/2019

Joint Segmentation and Path Classification of Curvilinear Structures

Detection of curvilinear structures in images has long been of interest....
research
11/21/2017

Identifying Most Walkable Direction for Navigation in an Outdoor Environment

We present an approach for identifying the most walkable direction for n...

Please sign up or login with your details

Forgot password? Click here to reset