A scalable, self‐supervised calibration and confounder removal model for opportunistic monitoring of road degradation

Hauwermeiren, Wout Van; Filipan, Karlo; Botteldooren, Dick; Coensel, Bert De

doi:10.1111/mice.12821

Cited by 3 publications

(8 citation statements)

References 34 publications

(52 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The source dataset has been thoroughly described in [2]. A brief introduction to the opportunistic roads vehicle dataset (ORVD2022): 41 vehicles of various types (sedans, light cargo vans, light passenger vans and light SUV's) collected data in various cities in Flanders from April 2020 until December 2021.…”

Section: A Opportunistic Sensing Of Roadsmentioning

confidence: 99%

“…In [2], a self-supervised calibration and confounder removal method (SCCR) had been proposed. An auto-encoding neural network is learned with a reconstruction loss term.…”

Section: B Self-supervised Calibration and Confounder Removalmentioning

confidence: 99%

“…As a consequence, no new sensors can be added after the model has been trained. As an alternative, a first effort had been made in [2] to characterize a sensor using a set of their observations and this approach is further explored in this paper.…”

Section: B Self-supervised Calibration and Confounder Removalmentioning

confidence: 99%

“…Then, the sensor readings can be rescaled to fit the reference signal. In [2], acoustic and vibration sensors have been deployed in multiple vehicles. The sensor devices are influenced by various internal and external factors such as vehicle construction, sensor installation, tire pressure, weight distribution of the vehicle, etc.…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, a more advanced calibration strategy is needed. To this end, a self calibration a self-calibration and confounder removal strategy (SCCR) for mobile road measurements is introduced [2]. SCCR is a deep learning auto-encoding neural network that reconstructs samples originating from one device, to another device under different circumstances.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Sensor characterization for supporting de-noising and auto-calibration of sensor data

Hauwermeiren

Hou

Filipan

et al. 2022

2022 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

In earlier work, a mobile sensor network had been deployed in passenger cars to measure road quality using sound and vibration sensors. A self-calibration and confounder removal (SCCR) algorithm eliminated the effect of the measurement device in sound and vibration measurements. It relied on the hypothesis that two measurements of the same object should give the same result. However, once the deep learning model has been trained, the model cannot easily include more devices, because it relies on a one-hot encoded device identification. Therefore, additional characterization of both sensor and vehicle is proposed here using a batch of observations (a bag) from the same sensors. A set-encoder is trained with a device classification head. This set-encoder compresses the input bag into a latent representation, irrespective of the input order. This latent vector, which characterizes the sensor, is injected into the SCCR framework to steer the calibration. New devices are characterized by the pre-trained set-encoder. This relies on the hypothesis that closely related sensors with a similar response should be close to each other in the latent space. Therefore, SCCR is now possible with new devices, previously unseen at the training time of SCCR. However, an increase in reconstruction error is observed for the unseen devices.

show abstract

Section: A Opportunistic Sensing Of Roadsmentioning

confidence: 99%

“…In [2], a self-supervised calibration and confounder removal method (SCCR) had been proposed. An auto-encoding neural network is learned with a reconstruction loss term.…”

Section: B Self-supervised Calibration and Confounder Removalmentioning

confidence: 99%

Section: B Self-supervised Calibration and Confounder Removalmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Sensor characterization for supporting de-noising and auto-calibration of sensor data

Hauwermeiren

Hou

Filipan

et al. 2022

2022 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

show abstract

Regeneration of pavement surface textures using M‐sigmoid‐normalized generative adversarial networks

Pan

Ren

et al. 2023

Computer aided Civil Eng

View full text Add to dashboard Cite

The high cost of collecting surficial textures has been the bottleneck problem for many decades. To bridge this gap, this study aims to propose a complete framework based on the generative adversarial networks (GANs) architecture for texture regeneration using limited texture data samples. Four variants of the GAN were compared to achieve the best regeneration performance. Prior to running GAN, specifically improved data augmentation and modified sigmoid normalization operations were dedicated to avoiding the loss of texture information and the non‐convergence problem. After running GAN, the estimation metrics were introduced and discussed separately to quantitatively evaluate the rationality and diversity of the regenerated textures. The results demonstrated that the use of improved data augmentation and normalization methods significantly enhanced the regeneration capability of GAN, and more importantly the skewed distribution of surficial heights which is not achievable by present texture regeneration methods was most effectively regenerated by deep regret analytic GAN with well generalization ability. Our study provides a promising choice to generate virtually realistic pavement surface textures.

show abstract

A deep learning framework based on improved self‐supervised learning for ground‐penetrating radar tunnel lining inspection

Huang

Yang

Zhou

et al. 2023

Computer aided Civil Eng

View full text Add to dashboard Cite

It is not practical to obtain a large number of labeled data to train a supervised learning network in tunnel lining nondestructive testing with ground‐penetrating radar (GPR). To decrease the dependence of supervised learning on the number of labeled data, an improved self‐supervised learning algorithm—self‐attention dense contrastive learning (SA‐DenseCL)—is proposed and incorporated with a mask region‐convolution neural network (Mask R‐CNN), which is trained by unlabeled and labeled GPR data. The proposed SA‐DenseCL adds a self‐attention‐based relevant projection head to the DenseCL architecture of self‐supervised learning, capturing the spatially continuing information between adjacent GPR traces. In the workflow, some unlabeled GPR images are used to pre‐train the SA‐DenseCL network for feature extraction and obtaining the backbone weights, which is superior to the conventional pre‐training methods of supervised learning pre‐trained by ImageNet images. The weights of the pre‐trained backbone are then used to initialize the Mask R‐CNN through transfer learning. Subsequently, a limited number of labeled GPR images are used to fine‐tune the Mask R‐CNN for automatically identifying the locations of the reinforcement bars and voids and estimating the secondary lining thickness. The experimental results show that the average precision reaches 96.70%, 81.04%, and 94.67% in identifying reinforcement bar locations, detecting void defects, and estimating secondary lining thickness, respectively, which outperform the conventional methods that use ImageNet‐based supervised learning or GPR image‐based DenseCL for initializing the Mask R‐CNN backbone weights. It is observed that the improved self‐supervised learning‐based framework can improve the detection and estimation accuracy in GPR tunnel lining inspection.

show abstract

A scalable, self‐supervised calibration and confounder removal model for opportunistic monitoring of road degradation

Cited by 3 publications

References 34 publications

Sensor characterization for supporting de-noising and auto-calibration of sensor data

Sensor characterization for supporting de-noising and auto-calibration of sensor data

Regeneration of pavement surface textures using M‐sigmoid‐normalized generative adversarial networks

A deep learning framework based on improved self‐supervised learning for ground‐penetrating radar tunnel lining inspection

Contact Info

Product

Resources

About