A Deep Learning-Based Approach for Road Surface Damage Detection

Kulambayev, Bakhytzhan; Beissenova, Gulbakhram; Katayev, Nazbek; Abduraimova, Bayan; Zhaidakbayeva, Lyazzat; Sarbassova, Alua; Akhmetova, Oxana; Issayev, Sapar; Suleimenova, Laura; Kasenov, Syrym; Shadinova, Kunsulu; Shyrakbayev, Abay

doi:10.32604/cmc.2022.029544

Cited by 7 publications

(7 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, the enhanced U-Net architecture's proficiency aligns with, and in certain respects surpasses, the capabilities of existing models. For instance, while previous studies using standard U-Net reported commendable performance [40], our model, with its integrative enhancements, demonstrated improved handling of the intricacies within pulmonary images. These enhancements, particularly the incorporation of attention gates, allowed for more nuanced feature recognition, addressing one of the primary limitations noted in past literature regarding convolutional neural networks' tendency for feature generalization [42].…”

Section: Discussionmentioning

confidence: 77%

“…Our current research into an enhanced U-Net architecture for lung CT segmentation synthesizes these collective insights and innovations, aiming to mitigate the existing challenges identified by previous studies. By integrating sophisticated context capture mechanisms, advanced convolution techniques, and a keen focus on model efficiency and interpretability [39][40], we contribute to the evolving landscape of AI-enhanced medical imaging. Through rigorous validation, we endeavor to underline the significance of our model in providing more accurate, reliable, and clinically applicable lung segmentation outputs, thereby influencing positive patient outcomes and resource optimization within healthcare systems [41].…”

Section: J Regulatory and Ethical Considerations In Ai-integrated Hea...mentioning

confidence: 99%

See 1 more Smart Citation

Enhanced U-Net Architecture for Lung Segmentation on Computed Tomography and X-Ray Images

Saimassay,

Begenov,

Sadyk

et al. 2024

ijacsa

View full text Add to dashboard Cite

In the expanding field of medical imaging, precise segmentation of anatomical structures is critical for accurate diagnosis and therapeutic interventions. This research paper introduces an innovative approach, building upon the established U-Net architecture, to enhance lung segmentation techniques applied to Computed Tomography (CT) images. Traditional methods of lung segmentation in CT scans often confront challenges such as heterogeneous tissue densities, variability in human anatomy, and pathological alterations, necessitating an approach that embodies greater robustness and precision. Our study presents a modified U-Net model, characterized by an integration of advanced convolutional layers and innovative skip connections, improving the reception field and facilitating the retention of high-frequency details essential for capturing the lung's intricate structures. The enhanced U-Net architecture demonstrates substantial improvements in dealing with the subtleties of lung parenchyma, effectively distinguishing between precarious nuances of tissues, and pathologies. Rigorous quantitative evaluations showcase a significant increase in the Dice coefficient and a decrease in the Hausdorff distance, indicating a more refined segmentation output compared to predecessor models. Additionally, the proposed model manifests exceptional versatility and computational efficiency, making it conducive for real-time clinical applications. This research underlines the transformative potential of employing advanced deep learning architectures for biomedical imaging, paving the way for early intervention, accurate diagnosis, and personalized treatment paradigms in pulmonary disorders. The findings have profound implications, propelling forward the nexus of artificial intelligence and healthcare towards unprecedented horizons.

show abstract

Section: Discussionmentioning

confidence: 77%

Section: J Regulatory and Ethical Considerations In Ai-integrated Hea...mentioning

confidence: 99%

Enhanced U-Net Architecture for Lung Segmentation on Computed Tomography and X-Ray Images

Saimassay,

Begenov,

Sadyk

et al. 2024

ijacsa

View full text Add to dashboard Cite

show abstract

“…Metrics such as accuracy, precision, recall, and F-score offer comprehensive assessments of model performance across various dimensions, including classification accuracy, false positive and false negative rates, and overall predictive capability [35]. The utilization of these metrics facilitates rigorous benchmarking against established standards and enables comparisons with prior research endeavors [36].…”

Section: Discussionmentioning

confidence: 99%

Development of Deep Learning Models for Traffic Sign Recognition in Autonomous Vehicles

Kozhamkulova,

Bidakhmet,

Vorogushina

et al. 2024

ijacsa

View full text Add to dashboard Cite

This research paper investigates the development of deep learning models for traffic sign recognition in autonomous vehicles. Leveraging convolutional neural networks (CNNs), the study explores various architectural configurations and evaluation methodologies to assess the efficacy of CNNs in accurately identifying and classifying traffic signs. Through a systematic evaluation process utilizing metrics such as accuracy, precision, recall, and F-score, the research demonstrates the robustness and generalization capability of the developed models across diverse environmental conditions. Furthermore, the utilization of visualization techniques, including the Matplotlib library, enhances the interpretability of model training dynamics and optimization progress. The findings highlight the significance of CNN architecture in facilitating hierarchical feature extraction and spatial dependency learning, thereby enabling reliable and efficient traffic sign recognition. The successful recognition of traffic signs under varying lighting conditions underscores the resilience of the developed models to environmental perturbations. Overall, this research contributes to advancing the capabilities of autonomous vehicle systems and lays the groundwork for the implementation of intelligent traffic sign recognition systems aimed at enhancing road safety and navigational efficiency.

show abstract

“…For instance, the approach proposed by [28] merely detects potholes under the D40 label, while Jana et al [29] differentiates damages strictly as longitudinal or transverse. Further, preceding deep learning studies [30][31][32][33] primarily focus on identifying the mere presence or absence of damage. www.ijacsa.thesai.org…”

Section: H Data Collection and Preparationmentioning

confidence: 99%

Real-Time Road Surface Damage Detection Framework based on Mask R-CNN Model

Kulambayev,

Nurlybek,

Astaubayeva

et al. 2023

IJACSA

View full text Add to dashboard Cite

In the ever-evolving realm of infrastructure management, the timely and accurate detection of road surface damages is imperative for the longevity and safety of transportation networks. This research paper introduces a pioneering framework centered on the Mask R-CNN (Regionbased Convolutional Neural Networks) model for real-time road surface damage detection. The overarching methodology encapsulates a deep learning-based approach to discern and classify various road aberrations such as potholes, cracks, and rutting. The chosen Mask R-CNN architecture, renowned for its proficiency in instance segmentation tasks, has been fine-tuned and optimized specifically for the unique challenges posed by road surfaces under diverse lighting and environmental conditions. A diverse dataset, amalgamating urban, suburban, and rural roadways under varied climatic conditions, served as the foundation for model training and validation. Preliminary results have not only underscored the model's robustness in realtime detection but also its superiority in terms of accuracy and computational efficiency when juxtaposed with extant methods. Concomitantly, the framework emphasizes scalability and adaptability, positing it as a frontrunner for potential integration into automated road maintenance systems and vehicular navigation aids. This trailblazing endeavor elucidates the potentialities of deep learning paradigms in revolutionizing road management systems, thus fostering safer and more efficient transportation environments.

show abstract

A Deep Learning-Based Approach for Road Surface Damage Detection

Cited by 7 publications

References 40 publications

Enhanced U-Net Architecture for Lung Segmentation on Computed Tomography and X-Ray Images

Enhanced U-Net Architecture for Lung Segmentation on Computed Tomography and X-Ray Images

Development of Deep Learning Models for Traffic Sign Recognition in Autonomous Vehicles

Real-Time Road Surface Damage Detection Framework based on Mask R-CNN Model

Contact Info

Product

Resources

About