Automatic Detection and Classification of Sewer Defects via Hierarchical Deep Learning

Xie, Qian; Li, Dawei; Xu, Jinxuan; Yu, Zhenghao; Wang, Jun

doi:10.1109/tase.2019.2900170

Cited by 144 publications

(59 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unlike the defects detection where the task is a binary classification problem, the defects usually involve multiple classes, and some models which can identify multiple defect types have been proposed. [4][5][6][7][8][9] Traditional detection methods mainly include manual detection, eddy current detection magnetic flux leakage detection and vision-based steel surface defect detection, but these methods usually lead to a large number of missing and false detections, which cannot guarantee product quality and meet the real-time need of the field. [10][11][12][13][14][15] Fortunately, deep learning methods have recently been applied to a number of similar domains with success.…”

Section: Surface Defects Classification Of Hot Rolled Strip Based On mentioning

confidence: 99%

Surface Defects Classification of Hot Rolled Strip Based on Improved Convolutional Neural Network

Wang

Lü

et al. 2021

ISIJ Int.

View full text Add to dashboard Cite

Section: Surface Defects Classification Of Hot Rolled Strip Based On mentioning

confidence: 99%

Surface Defects Classification of Hot Rolled Strip Based on Improved Convolutional Neural Network

Wang

Lü

et al. 2021

ISIJ Int.

View full text Add to dashboard Cite

“…However, along with the usage and aging of sewer systems, various sewer defects (e.g., cracks, obstacles and tree roots) have impacted the sewer conveyance capacities and brought up a series of socioeconomic and environmental problems [4,5]. Early detection of sewer defects provides support to system maintenance and management, and is thus an essential step towards improved sewer reliability [4,6,7]. Nevertheless, the traditional manual visual inspection of sewer defects is labor-intensive, subjective and error-prone, which can hardly meet the longterm development requirements of the large and complex modern sewer systems [5,8,9].…”

Section: Introductionmentioning

confidence: 99%

“…In recent years, increasing scholars have devoted attention to automated detection of sewer defects using deep learning techniques, such as the Convolutional Neural Networks (CNNs) [6,7], Faster R-CNN [11,12] and/or YOLO-based object detections [4,5]. Although these deep learning methods have showed great potential, they in general suffered from major drawbacks of limited training datasets.…”

Section: Introductionmentioning

confidence: 99%

Automated Sewer Defects Detection Using Style-Based Generative Adversarial Networks and Fine-Tuned Well-Known CNN Classifier

et al. 2021

View full text Add to dashboard Cite

Automated sewer defects detection has become an important trend for better management and maintenance of urban sewer systems. Deep learning technology has developed rapidly and offers an innovative solution for automated detection in engineering applications. However, insufficient data and unbalanced samples have proposed a big challenge to deep learning model training. This study adopts the state-of-the-art Style-based Generative Adversarial Networks (StyleGANs) model and compares the performances of its two variants in producing high-quality synthetic sewer defects images. Seven wellknown CNN models are further fine-tuned and trained using the synthetic images for automated sewer defects detection to examine the effects of StyleGANs on augmenting the detection performance. Results show that both StyleGANs are efficient in producing high-quality images with various styles and high-level details for multiple types of sewer defects. Specifically, the StyleGAN2-Adaptive Discriminator Augmentation (StyleGAN2-ADA) with the aid of Freeze Discriminator (Freeze-D) yields the best model performance. Among the adopted CNN classifiers, Inception_v3 achieves the highest detection accuracy. The mean detection accuracy is 94% (with a specific accuracy of 99.7%, 97%, 95.3% and 84% for tree root, residential wall, disjoint and obstacle, respectively) and confirms the reliability of the StyleGANs' performance. The study shows that StyleGANs provide a promising method to alleviate the limited and uneven dataset problem and can improve the deep learning model performance.

show abstract

“…These include customers’ sewer repair, replacement and rehabilitation, one of the major areas of spend [ 3 ]. This has opened an unprecedented opportunity for instrument makers and researchers to develop partnerships to offer novel approaches to proactively reduce the risk of failure and allow better understanding of the behaviour of the sewer network under both normal and extreme operating conditions [ 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 ].…”

Section: Introductionmentioning

confidence: 99%

“…Several fast and efficient alternative methods to analyse the condition of a sewer pipe wall objectively were proposed [ 8 , 10 , 18 , 19 , 20 , 21 , 22 , 23 ]. These methods are acoustic [ 8 , 10 , 18 ], acoustic optical [ 19 ], electromagnetic [ 20 , 21 ], electroacoustic [ 22 ], ultrasonic [ 5 , 7 , 23 ] and laser-based [ 8 ] methods. These methods are not designed to work in CSOs.…”

Section: Introductionmentioning

confidence: 99%

An Acoustic Sensor for Combined Sewer Overflow (CSO) Screen Condition Monitoring in a Drainage Infrastructure

See

Horoshenkov

Ali

et al. 2021

Sensors

View full text Add to dashboard Cite

Combined sewer overflow structures (CSO) play an important role in sewer networks. When the local capacity of a sewer system is exceeded during intense rainfall events, they act as a “safety valve” and discharge excess rainfall run-off and wastewater directly to a natural receiving water body, thus preventing widespread urban flooding. There is a regulatory requirement that solids in CSO spills must be small and their amount strictly controlled. Therefore, a vast majority of CSOs in the UK contain screens. This paper presents the results of a feasibility study of using low-cost, low-energy acoustic sensors to remotely assess the condition of CSO screens to move to cost-effective reactive maintenance visits. In situ trials were carried out in several CSOs to evaluate the performance of the acoustic sensor under realistic screen and flow conditions. The results demonstrate that the system is robust within ±2.5% to work successfully in a live CSO environment. The observed changes in the screen condition resulted in 8–39% changes in the values of the coefficient in the proposed acoustic model. These changes are detectable and consistent with observed screen and hydraulic data. This study suggested that acoustic-based sensing can effectively monitor the CSO screen blockage conditions and hence reduce the risk of non-compliant CSO spills.

show abstract

Automatic Detection and Classification of Sewer Defects via Hierarchical Deep Learning

Cited by 144 publications

References 37 publications

Surface Defects Classification of Hot Rolled Strip Based on Improved Convolutional Neural Network

Surface Defects Classification of Hot Rolled Strip Based on Improved Convolutional Neural Network

Automated Sewer Defects Detection Using Style-Based Generative Adversarial Networks and Fine-Tuned Well-Known CNN Classifier

An Acoustic Sensor for Combined Sewer Overflow (CSO) Screen Condition Monitoring in a Drainage Infrastructure

Contact Info

Product

Resources

About