What are the most common machine learning algorithms used for damage detection in asphalts and roads?
Best insight from top research papers
Machine learning algorithms commonly used for damage detection in asphalt and roads include Artificial Neural Networks (ANN), Support Vector Machine (SVM), Convolutional Neural Networks (CNN), and ResNet-34. Marulanda and Thomson proposed a methodology that applied pattern recognition and image analysis with ANN and SVM for the detection of potholes, longitudinal and alligator cracks . Gupta, Chauhan, and Singla implemented CNN and ResNet-34 to detect road damage, achieving accuracies of 94.79% and 89.94% respectively . Liu, Wang, Luo, Lv, Guo, and Xie utilized machine learning and INS data to detect pavement damage, achieving an F1 score of 99.61% and precision of 99.33% .
Answers from top 5 papers
More filters
| Papers (5) | Insight |
|---|---|
11 May 2023 | The most common machine learning algorithms used for damage detection in asphalts and roads are Convolutional Neural Network (CNN) and ResNet-34. |
| The most common machine learning algorithms used for damage detection in roads are Convolutional Neural Networks (CNN), R-CNN, and Faster R-CNN. | |
| The paper does not mention the specific machine learning algorithms used for damage detection in asphalts and roads. | |
| The most common machine learning algorithms used for damage detection in asphalts and roads are Artificial Neural Networks (ANN) and Support Vector Machine (SVM). | |
11 May 2023 | The most common machine learning algorithms used for damage detection in asphalts and roads are Convolutional Neural Network (CNN) and ResNet-34. |
Related Questions
What are the most common traditional methods used in automatic pavement crack detection?5 answersTraditional methods commonly used in automatic pavement crack detection include image segmentation techniques with convolutional neural networks (CNNs) for pixel-level crack identification. These methods often involve manual predetermination of crack locations, image selection, and segmentation processes, which can be labor-intensive and prone to errors. Additionally, visible light sensors have been traditionally utilized for crack detection, but they are sensitive to lighting conditions, leading to false positives or missed detections in low-light environments. To address these challenges, some traditional approaches incorporate post-processing techniques like connected components labeling algorithms to improve segmentation accuracy and reduce interference from irrelevant non-crack regions in pavement images. These methods have been effective but are being enhanced by newer, more advanced techniques for improved accuracy and efficiency in crack detection.
Best deep learning algorithms be used to detect road potholes?5 answersThe best deep learning algorithms for road pothole detection include YOLOv4, YOLOv4-tiny, and YOLOv5s for their high accuracy and efficiency in real-time applications. Additionally, a study comparing Sigmoid and Softmax activation functions in CNN models found that Sigmoid outperformed Softmax in detecting potholes, achieving higher accuracy scores across different datasets. Moreover, a model utilizing CNN and ResNet architectures achieved accuracies of 94.79% and 89.94% respectively in detecting various road damages, including potholes, demonstrating the effectiveness of these algorithms in pavement defect detection. Overall, these deep learning models offer robust solutions for accurately identifying road potholes, crucial for enhancing road safety and maintenance.
Best deep learning algorithms be used to detect road cracks?5 answersThe best deep learning algorithms for detecting road cracks include Xception, CNN, ResNet, and SwinUNet. Xception has shown high accuracy and predictive performance, making it a valuable tool for image classification and object recognition. CNN and ResNet have also been used successfully, achieving accuracies of 94.79% and 89.94% respectively. SwinUNet, a transformer-based model, outperforms other transformers and shows the highest accuracy among nine models tested. These deep learning algorithms have been proven effective in detecting surface cracks in pavement, providing valuable insights for future applications in this field.
Why machine learning models used to identify the significant factors that effect on road accidents?5 answersMachine learning models are used to identify the significant factors that affect road accidents because they can analyze large amounts of data from various sources such as accident reports, weather conditions, and road infrastructure. These models can use supervised learning algorithms like Decision Tree, Naive Bayes, Random Forest, K-Nearest Neighbor, and Logistic Regression to predict the severity of accidents and understand the relationships between accidents, road conditions, and environmental factors. By analyzing these factors, governments and researchers can develop strategies to reduce the occurrence of accidents and mitigate their associated risks. Additionally, machine learning models can provide real-time risk information to road users, helping them make informed decisions and avoid potential accidents. These models can also be used to improve the accuracy of accident reporting systems and facilitate communication with relevant authorities such as hospitals, ambulances, and insurance companies.
Are the best models using traditional methodology or are neural networks better for detecting potholes on highways?4 answersNeural networks, specifically deep learning algorithms such as convolutional neural networks (CNNs), have shown promise in detecting potholes on highways. These algorithms are designed to process images and extract relevant patterns and characteristics, making them suitable for image recognition tasks like pothole detection. Traditional methodologies, on the other hand, involve techniques like computer vision and supervised learning, where features are extracted from images using methods like Gray-Level Co-Occurrence Matrix (GLCM), Local Binary Pattern (LBP), and Gabor Filter's texture descriptors, and classification models like Support Vector Machines (SVM), k-Nearest Neighbors (kNN), and Multi-Layer Perceptron (MLP) are used. While both approaches have been explored, the use of neural networks, particularly CNNs, has shown excellent accuracy in various image recognition applications, making them a promising choice for pothole detection on highways.
Review the current techniques on vibration-based damage identification such as machine learning, deep learning?4 answersVibration-based damage identification techniques, such as machine learning (ML) and deep learning (DL), have gained popularity in recent years. ML methods can be used for pre- and post-processing input data, extracting damage sensitive features, and operating pattern recognition in measured data. These techniques utilize vibration monitoring hardware and software to measure and process acceleration, velocity, and displacement-based signals, which are then analyzed to correlate with the presence of damages. ML algorithms have significantly enhanced the capabilities of structural health monitoring (SHM) systems, providing innovative solutions to challenges in damage detection and monitoring. Additionally, the performance of ML approaches has been compared to early-developed damage indicators, such as frequency and mode shapes, showing promising results. Furthermore, DL algorithms have become more feasible and extensively used in vibration-based structural damage detection, offering elegant performance and accuracy.