Joint Learning of Blind Super-Resolution and Crack Segmentation for Realistic Degraded Images (2023-02-24T00:00:00.000000Z)

TL;DR

This article proposes crack segmentation augmented by super-resolution (SR) with deep neural networks, and proposes two extra paths that further encourage the mutual optimization between SR and segmentation.

Abstract

This article proposes crack segmentation augmented by super-resolution (SR) with deep neural networks. In the proposed method, an SR network is jointly trained with a binary segmentation network in an end-to-end manner. This joint learning allows the SR network to be optimized for improving segmentation results. For realistic scenarios, the SR network is extended from nonblind to blind for processing a low-resolution (LR) image degraded by unknown blurs. The joint network is improved by our proposed two extra paths that further encourage the mutual optimization between SR and segmentation. Comparative experiments with state-of-the-art (SoTA) segmentation methods demonstrate the superiority of our joint learning, and various ablation studies prove the effects of our contributions.

Authors

Yuki Kondo

4 papers

N. Ukita

7 papers

TL;DR

Abstract

Authors

References130 items

Kernelized Back-Projection Networks for Blind Super Resolution

Classification of cervical cells leveraging simultaneous super-resolution and ordinal regression

Embryo Grading With Unreliable Labels Due to Chromosome Abnormalities by Regularized PU Learning With Ranking

Dual Focal Loss to address class imbalance in semantic segmentation

Erroneous pixel prediction for semantic image segmentation

CrackFormer: Transformer Network for Fine-Grained Crack Detection

Developing a new deep learning CNN model to detect and classify highway cracks

Distant Bird Detection for Safe Drone Flight and Its Dataset

Crack Segmentation for Low-Resolution Images using Joint Learning with Super- Resolution

AVM Image Quality Enhancement by Synthetic Image Learning for Supervised Deblurring

Automatic crack segmentation using deep high-resolution representation learning.

MASA-SR: Matching Acceleration and Spatial Adaptation for Reference-Based Image Super-Resolution

NTIRE 2021 Challenge on Burst Super-Resolution: Methods and Results

Learning the Non-differentiable Optimization for Blind Super-Resolution

Automatic crack classification and segmentation on masonry surfaces using convolutional neural networks and transfer learning

Unsupervised Degradation Representation Learning for Blind Super-Resolution

Designing a Practical Degradation Model for Deep Blind Image Super-Resolution

Loss odyssey in medical image segmentation

Learning to Resize Images for Computer Vision Tasks

PointFlow: Flowing Semantics Through Points for Aerial Image Segmentation

Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation

Analyzing Overfitting Under Class Imbalance in Neural Networks for Image Segmentation

KOALAnet: Blind Super-Resolution using Kernel-Oriented Adaptive Local Adjustment

Region-dependent Scale Proposals for Super-Resolution in Object Detection

Comparison of crack segmentation using digital image correlation measurements and deep learning

Unfolding the Alternating Optimization for Blind Super Resolution

AIM 2020 Challenge on Video Extreme Super-Resolution: Methods and Results

SDDNet: Real-Time Crack Segmentation

Correction to: Single Image Super-Resolution via a Holistic Attention Network

KGSNet: Key-Point-Guided Super-Resolution Network for Pedestrian Detection in the Wild

CrackGAN: Pavement Crack Detection Using Partially Accurate Ground Truths Based on Generative Adversarial Learning

SRFlow: Learning the Super-Resolution Space with Normalizing Flow

Dual Super-Resolution Learning for Semantic Segmentation

Robust Reference-Based Super-Resolution With Similarity-Aware Deformable Convolution

Multi-Image Super-Resolution for Remote Sensing using Deep Recurrent Networks

FBRNN: feedback recurrent neural network for extreme image super-resolution

Deep super resolution crack network (SrcNet) for improving computer vision–based automated crack detectability in in situ bridges

NTIRE 2020 Challenge on Perceptual Extreme Super-Resolution: Methods and Results

Space-Time-Aware Multi-Resolution Video Enhancement

Deep Unfolding Network for Image Super-Resolution

Closed-Loop Matters: Dual Regression Networks for Single Image Super-Resolution

Stochastic Frequency Masking to Improve Super-Resolution and Denoising Networks

Meta-Transfer Learning for Zero-Shot Super-Resolution

Image Segmentation Using Deep Learning: A Survey

Fast Adaptation to Super-Resolution Networks via Meta-Learning

Pixel-Level Cracking Detection on 3D Asphalt Pavement Images Through Deep-Learning- Based CrackNet-V

Explorable Super Resolution

Correction Filter for Single Image Super-Resolution: Robustifying Off-the-Shelf Deep Super-Resolvers

Blind image super-resolution with spatially variant degradations

AIM 2019 Challenge on Image Extreme Super-Resolution: Methods and Results

Kernel Modeling Super-Resolution on Real Low-Resolution Images

Object-Contextual Representations for Semantic Segmentation

Anchor Loss: Modulating Loss Scale Based on Prediction Difficulty

Dual Directed Capsule Network for Very Low Resolution Image Recognition

NTIRE 2019 Challenge on Video Super-Resolution: Methods and Results

Geometry-Aware Distillation for Indoor Semantic Segmentation

Second-Order Attention Network for Single Image Super-Resolution

JCS-Net: Joint Classification and Super-Resolution Network for Small-Scale Pedestrian Detection in Surveillance Images

SinGAN: Learning a Generative Model From a Single Natural Image

Reducing the Hausdorff Distance in Medical Image Segmentation With Convolutional Neural Networks

Blind Super-Resolution With Iterative Kernel Correction

Deep Back-ProjectiNetworks for Single Image Super-Resolution

DeepCrack: A deep hierarchical feature learning architecture for crack segmentation

Deep Plug-And-Play Super-Resolution for Arbitrary Blur Kernels

Recurrent Back-Projection Network for Video Super-Resolution

Feedback Network for Image Super-Resolution

Image Super-Resolution by Neural Texture Transfer

DeepCrack: Learning Hierarchical Convolutional Features for Crack Detection

Deep Learning for Image Super-Resolution: A Survey

Feature Pyramid and Hierarchical Boosting Network for Pavement Crack Detection

Boundary loss for highly unbalanced segmentation

Tunnel inspection using photogrammetric techniques and image processing: A review

2018 PIRM Challenge on Perceptual Image Super-resolution

SOD-MTGAN: Small Object Detection via Multi-Task Generative Adversarial Network

Deep Learning-Based Fully Automated Pavement Crack Detection on 3D Asphalt Surfaces with an Improved CrackNet

Sharpness Enhancement and Super-Resolution of Around-View Monitor Images