• Home
  • Search
  • Improved Noise and Attack Robustness for Semantic Segmentation by Using Multi-Task Training with Self-Supervised Depth Estimation
  • Open Access IconOpen Access
  • Cite Icon51
  • https://doi.org/10.1109/cvprw50498.2020.00168Copy DOI Icon

Improved Noise and Attack Robustness for Semantic Segmentation by Using Multi-Task Training with Self-Supervised Depth Estimation

  • Jun 1, 2020
  • Marvin Klingner +2 more
Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

While current approaches for neural network training often aim at improving performance, less focus is put on training methods aiming at robustness towards varying noise conditions or directed attacks by adversarial examples. In this paper, we propose to improve robustness by a multi-task training, which extends supervised semantic segmentation by a self-supervised monocular depth estimation on unlabeled videos. This additional task is only performed during training to improve the semantic segmentation model's robustness at test time under several input perturbations. Moreover, we even find that our joint training approach also improves the performance of the model on the original (supervised) semantic segmentation task. Our evaluation exhibits a particular novelty in that it allows to mutually compare the effect of input noises and adversarial attacks on the robustness of the semantic segmentation. We show the effectiveness of our method on the Cityscapes dataset, where our multi-task training approach consistently outperforms the single-task semantic segmentation baseline in terms of both robustness vs. noise and in terms of adversarial attacks, without the need for depth labels in training.

Similar Papers
  • Book Chapter
  • Citations44

Chapter 12 - Semantic scene segmentation for robotics

  • Jan 01, 2022
  • Deep Learning for Robot Perception and Cognition
  • Juana Valeria Hurtado +1
  • Conference Article
  • Citations7

Learning Panoptic Segmentation from Instance Contours

  • May 30, 2021
  • Sumanth Chennupati +4
  • Conference Article
  • Citations10

Application of shared backbone DNNs in ADAS perception systems

  • Jan 04, 2021
  • Mikhail G Lobanov +1
  • Conference Article
  • Citations27

Continual Learning for Class- and Domain-Incremental Semantic Segmentation

  • Jul 11, 2021
  • Tobias Kalb +3
  • Research Article
  • Citations12

SENSE: Self-Evolving Learning for Self-Supervised Monocular Depth Estimation.

  • Jan 01, 2024
  • IEEE Transactions on Image Processing
  • Guanbin Li +4
  • Book Chapter

Transferring Styles for Reduced Texture Bias and Improved Robustness in Semantic Segmentation Networks

  • Oct 21, 2025
  • Frontiers in artificial intelligence and applications
  • Ben Hamscher +4
  • Conference Article
  • Citations8

Not All Pixels are Born Equal

  • Oct 15, 2018
  • Vikash Sehwag +5
  • Book Chapter
  • Citations87

Characterizing Adversarial Examples Based on Spatial Consistency Information for Semantic Segmentation

  • Jan 01, 2018
  • Chaowei Xiao +5
  • Conference Article

A Novel Framework for Semantic Image Segmentation using Generative Adversarial Networks and Neural Architecture Search

  • Aug 11, 2021
  • Vayangi Vishmi Vishara Ganepola +1
  • Conference Article

Joint Depth Estimation and Semantic Segmentation with Adversarial Multi-task Network

  • Oct 01, 2020
  • Hui Ren +2
  • Conference Article
  • Citations32

Semantic Segmentation With Multi Scale Spatial Attention For Self Driving Cars

  • Oct 01, 2021
  • Abhinav Sagar +1
  • Conference Article
  • Citations20

Semantic Segmentation of Autonomous Driving Images by the Combination of Deep Learning and Classical Segmentation

  • Mar 03, 2021
  • Mohammad Hosein Hamian +3
  • Research Article
  • Citations2

Unmanned aerial vehicle visual scene understanding based on multitask learning network

  • May 03, 2023
  • Journal of Electronic Imaging
  • Zhicheng Nie +3
  • Conference Article

Unsupervised Domain Adaptation via Shared Content Representation for Semantic Segmentation

  • Sep 19, 2021
  • Christian Hubschneider +2
  • Research Article
  • Citations74

Exploiting Depth From Single Monocular Images for Object Detection and Semantic Segmentation.

  • Oct 26, 2016
  • IEEE Transactions on Image Processing
  • Yuanzhouhan Cao +2
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.