Detecting excessive load-carrying tasks using a deep learning network with a Gramian Angular Field

Lee, Hoonyong; Yang, Keun–Hyeok; Kim, Namgyun

doi:10.1016/j.autcon.2020.103390

Cited by 50 publications

(22 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As both outcomes confirm with each other, we follow the relevance's trend and ensure that our research is based on domains that manifest a high potential to maintain or improve their importance. Furthermore, related work in time-series imaging and corresponding analysis using DL-based techniques reveals that our identified domains for benchmarking investigation (i.e., medicine and engineering) confirm with prior research in the field [32], [38], [57], [58].…”

Section: A Dim1: Data Foundationsupporting

confidence: 79%

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

2022

View full text Add to dashboard Cite

With the growing availability and complexity of time-series sequences, scalable and robust machine learning approaches are required that overcome the sampling challenge of quantitatively sufficient training data. Following the research trend towards the deep learning-based analysis of time-series encoded as images, this study proposes a time-series imaging workflow that overcomes the challenge of quantitatively limited sensor data across domains (i.e., medicine and engineering). After systematically identifying the three relevant dimensions that affect the performance of the deep learning-based analysis of visualized timeseries data, we performed a benchmarking evaluation with a total of 24 unique convolutional neural network models. Following a two-level transfer learning investigation, we reveal that fine-tuning the mid-level features results in the best classification performance. As a result, we present an optimized representation of the VGG16 network, which outperforms previous studies in the field. Our approach is accurate, robust, and manifests internal and external validity. By only using the raw time-series data, our model does not require manual feature engineering, being of high practical relevance. As the post-hoc analysis of our results reveals that our model allows automated extraction of meaningful features based on the trend of the underlying timeseries data, our study also adds to explainable artificial intelligence. Furthermore, our proposed workflow reduces the sequence length of the input data while preserving all information. Especially with the hurdle of long-term dependencies in sequential time-series data, we overcome related work's limitation of the vanishing gradients problem and contribute to the sequential learning theory in artificial intelligence.

show abstract

Section: A Dim1: Data Foundationsupporting

confidence: 79%

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

2022

View full text Add to dashboard Cite

show abstract

“…On the other hand, Lee et al [10] have addressed the prediction of Work-Related Musculoskeletal Disorders (WMSDs) while excessive load carrying during various construction tasks. In this study, the GAF transform was employed to convert data coming from the inertial measurement unit sensors to into images before performing a hybrid CNN-LSTM to classify load-carrying modes.…”

Section: Related Workmentioning

confidence: 99%

Elevating Energy Data Analysis with M2GAF: Micro-Moment Driven Gramian Angular Field Visualizations

Alsalemi¹,

Amira²,

Malekmohamadi³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Shi et al (2020) has proposed deep LSTM network to predict the brake pedal aperture for different braking types. Lee et al (2020) have proposed an automatic detecting technique for excessive carrying-load (DeTECLoad) uses a hybrid Convolutional Neural Network-Long Short-Term Memory to predict load-carrying weights and postures simultaneously.…”

Section: Health and Safety Using Deep Learningmentioning

confidence: 99%

“…Shi et al (2020) has proposed deep LSTM network to predict the brake pedal aperture for different braking types. Lee et al (2020) Convolutional Neural Network-Long Short-Term Memory to predict load-carrying weights and postures simultaneously. 5.2.4 Personal protective equipment's.…”

Section: Construction Machines Detectionsmentioning

confidence: 99%

The application of “deep learning” in construction site management: scientometric, thematic and critical analysis

Elghaish

Matarneh

Alhusban

2021

View full text Add to dashboard Cite

Purpose The digital construction transformation requires using emerging digital technology such as deep learning to automate implementing tasks. Therefore, this paper aims to evaluate the current state of using deep learning in the construction management tasks to enable researchers to determine the capabilities of current solutions, as well as finding research gaps to carry out more research to bridge revealed knowledge and practice gaps. Design/methodology/approach The scientometric analysis is conducted for 181 articles to assess the density of publications in different topics of deep learning-based construction management applications. After that, a thematic and gap analysis are conducted to analyze contributions and limitations of key published articles in each area of application. Findings The scientometric analysis indicates that there are four main applications of deep learning in construction management, namely, automating progress monitoring, automating safety warning for workers, managing construction equipment, integrating Internet of things with deep learning to automatically collect data from the site. The thematic and gap analysis refers to many successful cases of using deep learning in automating site management tasks; however, more validations are recommended to test developed solutions, as well as additional research is required to consider practitioners and workers perspectives to implement existing applications in their daily tasks. Practical implications This paper enables researchers to directly find the research gaps in the existing solutions and develop more workable applications to bridge revealed gaps. Accordingly, this will be reflected on speeding the digital construction transformation, which is a strategy over the world. Originality/value To the best of the authors’ knowledge, this paper is the first of its kind to adopt a structured technique to assess deep learning-based construction site management applications to enable researcher/practitioners to either adopting these applications in their projects or conducting further research to extend existing solutions and bridging revealed knowledge gaps.

show abstract

Detecting excessive load-carrying tasks using a deep learning network with a Gramian Angular Field

Cited by 50 publications

References 59 publications

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

Elevating Energy Data Analysis with M2GAF: Micro-Moment Driven Gramian Angular Field Visualizations

The application of “deep learning” in construction site management: scientometric, thematic and critical analysis

Contact Info

Product

Resources

About