Hybrid Skip: A Biologically Inspired Skip Connection for the UNet Architecture

07/11/2022
by   Nikolaos Zioulis, et al.
15

In this work we introduce a biologically inspired long-range skip connection for the UNet architecture that relies on the perceptual illusion of hybrid images, being images that simultaneously encode two images. The fusion of early encoder features with deeper decoder ones allows UNet models to produce finer-grained dense predictions. While proven in segmentation tasks, the network's benefits are down-weighted for dense regression tasks as these long-range skip connections additionally result in texture transfer artifacts. Specifically for depth estimation, this hurts smoothness and introduces false positive edges which are detrimental to the task due to the depth maps' piece-wise smooth nature. The proposed HybridSkip connections show improved performance in balancing the trade-off between edge preservation, and the minimization of texture transfer artifacts that hurt smoothness. This is achieved by the proper and balanced exchange of information that Hybrid-Skip connections offer between the high and low frequency, encoder and decoder features, respectively.

READ FULL TEXT

page 1

page 3

page 6

page 9

page 10

research
08/29/2022

Rethinking Skip Connections in Encoder-decoder Networks for Monocular Depth Estimation

Skip connections are fundamental units in encoder-decoder networks, whic...
research
10/17/2022

Attention Attention Everywhere: Monocular Depth Prediction with Skip Attention

Monocular Depth Estimation (MDE) aims to predict pixel-wise depth given ...
research
10/10/2020

Selective Information Passing for MR/CT Image Segmentation

Automated medical image segmentation plays an important role in many cli...
research
05/24/2022

UNet#: A UNet-like Redesigning Skip Connections for Medical Image Segmentation

As an essential prerequisite for developing a medical intelligent assist...
research
01/05/2021

Novel View Synthesis via Depth-guided Skip Connections

We introduce a principled approach for synthesizing new views of a scene...
research
02/10/2021

Sequential vessel segmentation via deep channel attention network

This paper develops a novel encoder-decoder deep network architecture wh...
research
01/02/2022

Recurrent Feature Propagation and Edge Skip-Connections for Automatic Abdominal Organ Segmentation

Automatic segmentation of abdominal organs in computed tomography (CT) i...

Please sign up or login with your details

Forgot password? Click here to reset