Damage detection using in-domain and cross-domain transfer learning

Bukhsh, Zaharah Allah; Jansen, Nils; Saeed, Aaqib

doi:10.1007/s00521-021-06279-x

Cited by 33 publications

(9 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, in Bukhsh et al. (2021), in‐domain and cross‐domain TL is analyzed for damage classification. The analysis shows that both types of TL enhance the performance of networks, especially on small data sets.…”

Section: Related Workmentioning

confidence: 99%

A sigmoid‐optimized encoder–decoder network for crack segmentation with copy‐edit‐paste transfer learning

Çelik

König

2022

Computer aided Civil Eng

View full text Add to dashboard Cite

The automatic recognition of cracks is an essential requirement for the cost‐efficient maintenance of concrete structures, such as bridges, buildings, and roads. It should allow the localization and the determination of the crack type and the evaluation of the crack severity by providing information on the shape, orientation, and crack area and width. The first step in this direction is the automatized segmentation of cracks. This paper provides a concrete crack data set (370 images) and proposes two solutions that achieve the best results on two different crack data sets. Our first solution concerns the segmentation architecture. We provide an encoder–decoder‐based network with a particular interconnection of layers between the encoder and decoder parts that outperforms several other methods. In addition, this network is enhanced by squeeze‐and‐excitation blocks equipped with a modified sigmoid activation function. We introduce a stretch coefficient into the sigmoid function and declare it a trainable parameter, allowing more differentiated calibration of the feature map during network training. Our second solution concerns kernel initialization by transfer learning (TL). We propose the Copy‐Edit‐Paste Transfer Learning (CEP TL). By copying, geometric editing, and pasting crack masks onto new concrete background images, we generate thousands of semisynthetic images used to pretrain the network. This CEP TL method increases model performance with significant differences. For data set A (ours), we achieve F1‐scores 76.06 ± 0.06% without CEP TL and 92.32 ± 0.82% with CEP TL. For data set B (DeepCrack data set), we achieve F1‐scores 88.56 ± 0.01% without CEP TL and 90.59 ± 0.80% with CEP TL.

show abstract

Section: Related Workmentioning

confidence: 99%

A sigmoid‐optimized encoder–decoder network for crack segmentation with copy‐edit‐paste transfer learning

Çelik

König

2022

Computer aided Civil Eng

View full text Add to dashboard Cite

show abstract

“… 31 Second, the source data set and the target data set must have the same or similar distribution. 32 Therefore, it takes a lot of computing resources to simulate the sample library. In addition, the number of parameters that need to be trained for the stacked LSTM network is very large, and the training process consumes a lot of time and computing resources.…”

Section: Transfer Learning Based On the Pretraining Modelmentioning

confidence: 99%

“…First, the source data set must be labeled sample data . Second, the source data set and the target data set must have the same or similar distribution . Therefore, it takes a lot of computing resources to simulate the sample library.…”

Section: Transfer Learning Based On the Pretraining Modelmentioning

confidence: 99%

Reservoir Production Prediction Model Based on a Stacked LSTM Network and Transfer Learning

et al. 2021

View full text Add to dashboard Cite

Gas injection and water injection are common and effective methods to improve oil recovery. To ensure its production effect, it is necessary to simulate the oilfield production process. However, traditional composition simulation runs a large number of calculations and takes a long time. Through the analysis of relevant data, we found that production is affected by many factors and has a strong sequential character. Therefore, this paper proposes a deep learning model for reservoir production prediction based on stacked long short-term memory network (LSTM). It is applied to other well patterns with a short production time and a few samples in the same oilfield block by transfer learning. The model achieves an effective combination with the actual reservoir production process. At the same time, it uses the knowledge learned from the well pattern with sufficient historical data to assist in the establishment of the model of the well pattern with limited data. This can obtain accurate prediction results and save the model training time, thus getting more effective application effects than composition simulation. This paper verifies the effectiveness of the proposed method through the data and multiple different injection combinations of the Tarim oilfield.

show abstract

“…Xia et al [10] proposed a graph alignment mechanism for mapping source and target-domain information for Unsupervised Domain Adaptation (UDA) with insufficient or unlabelled data. Further, Buksh et al [11] used a mix of indomain and cross-domain learning on tiny image datasets to successfully detect structural damage in bridges. Similarly, Lin et al [12] used cross-domain learning for designing a damage-sensitive and domain-invariant feature extractor for structural damage detection.…”

Section: Introductionmentioning

confidence: 99%

Ubiquitous Domain Adaptation at the Edge for Vibration-Based Machine Status Monitoring

Mukherjee

Parlikad

McFarlane

2023

IEEE Internet Things J.

View full text Add to dashboard Cite

This paper presents a Ubiquitous Domain Adaptation (UDA) and generalizability technique for vibrationbased automated machine status monitoring at the edge. The method significantly reduces the effects of signal noise artifacts and device/usage-specific vibration signatures using basic timefrequency domain signal operations and a lightweight ensemble of data-driven classifiers, allowing the method to be used for reliable domain-invariant status monitoring of motorized equipment. An experimental setup using vibration data from an air-cooled electric blender motor (source-domain) is used to train an automated machine state identification classifier that can identify the operating states of an eccentric rotating mass vibration motor (target-domain). Initial deployment of this method on target-domain motorized devices resulted in a machine status monitoring accuracy of at least 81.6% and a maximum training accuracy of almost 99% on known data of the source-domain and 91.49% for unseen data in the target-domain within an acceptable time frame. The performance of the proposed method is also comparable across platforms ranging from resource-constrained edge to a resource-rich cloud. This approach facilitates the use of noisy or uncalibrated sensor data in data-driven machine status monitoring tasks, therefore allowing for the development of reusable, low-cost monitoring systems that require meagre developmental effort, resulting in accelerated deployment times.

show abstract

Damage detection using in-domain and cross-domain transfer learning

Cited by 33 publications

References 59 publications

A sigmoid‐optimized encoder–decoder network for crack segmentation with copy‐edit‐paste transfer learning

A sigmoid‐optimized encoder–decoder network for crack segmentation with copy‐edit‐paste transfer learning

Reservoir Production Prediction Model Based on a Stacked LSTM Network and Transfer Learning

Ubiquitous Domain Adaptation at the Edge for Vibration-Based Machine Status Monitoring

Contact Info

Product

Resources

About