RoadCare: A Deep-learning Based Approach to Quantifying Road Surface Quality
TL;DR: This paper proposes a deep-learning based approach to road surface quality monitoring, using accelerometer and GPS sensor readings, which enables several useful smart-city applications such as spatio-temporal monitoring of the city's roads, early warning of bad road conditions, as well as choosing the "smoothest" road route to a destination.
Abstract: Roads form a critical part of any region's infrastructure. Their constant monitoring and maintenance is thus essential. Traditional monitoring mechanisms are heavy-weight, and hence have insufficient coverage. In this paper, we explore the use of crowd-sourced intelligent measurements from commuters' smart-phone sensors. Specifically, we propose a deep-learning based approach to road surface quality monitoring, using accelerometer and GPS sensor readings. Through extensive data collection of over 36 hours on different kinds of roads, and subsequent evaluation based on this, we show that the approach can achieve high accuracy (98.5%) in a three-way classification of road surface quality. We also show how the classification can be extended to a finer grained 11-point scale of road quality. The model is also efficient: it can be implemented on today's smart-phones, thus making it practical. Our approach, called RoadCare, enables several useful smart-city applications such as spatio-temporal monitoring of the city's roads, early warning of bad road conditions, as well as choosing the "smoothest" road route to a destination.
Citations
More filters
Journal Article•
TL;DR: Understanding of the occurrence and development of road traffic injuries will contribute to the prevention and control of crash and to the implementation of "everybody has the right to enjoy health" proposed by WHO.
Abstract: The appearance of cars has raised materialistic civilization and living standard to an unprecedented level. Today, it is hard to imagine how we human beings can live without cars. Yet, motor vehicles can cause a great number of deaths and injuries as well as considerable economic losses, which have constituted the global burden. Understanding of the occurrence and development of road traffic injuries will contribute to the prevention and control of crash and to the implementation of "everybody has the right to enjoy health" proposed by WHO.
312 citations
TL;DR: In this article, the authors proposed a short survey on the use of smartphone sensors in the detection of various kinds of anomalies in several fields namely environment, agriculture, healthcare and road/traffic conditions.
Abstract: With smartphones being so ubiquitous, more connected and largely fitted with several types of sensors such as GPS, microphones, cameras, magnetometers, accelerometers, etc; there is an increasing opportunity in the development of smartphone-based sensor systems. In this article, we propose a short survey on the use of smartphone sensors in the detection of various kinds of anomalies in several fields namely environment, agriculture, healthcare and road/traffic conditions. We also list the main advantages and limitations of the use of smartphone sensors systems in such fields.
36 citations
TL;DR: In this article , the authors describe the body of knowledge in smartphone-based roughness assessment, report knowledge gaps and cast light on future research directions. But, they focus on practical factors that are expected to affect the accuracy and robustness of smartphonebased methods, including data collection speed, vehicle type, smartphone specifications and mounting configuration.
11 citations
01 Mar 2018
8 citations
TL;DR: The results of this review show that road surface anomaly detection and classification performed through vibration-based methods have achieved relatively high performance, however, there are challenges related to the reproduction and heterogeneity of the results that have been reported that are influenced by the limited testing conditions, sample size, and lack of publicly available datasets.
Abstract: Road surfaces suffer from sources of deterioration, such as weather conditions, constant usage, loads, and the age of the infrastructure. These sources of decay generate anomalies that could cause harm to vehicle users and pedestrians and also develop a high cost to repair the irregularities. These drawbacks have motivated the development of systems that automatically detect and classify road anomalies. This study presents a narrative review focused on road surface anomaly detection and classification based on vibration-based techniques. Three methodologies were surveyed: threshold-based methods, feature extraction techniques, and deep learning techniques. Furthermore, datasets, signals, preprocessing steps, and feature extraction techniques are also presented. The results of this review show that road surface anomaly detection and classification performed through vibration-based methods have achieved relatively high performance. However, there are challenges related to the reproduction and heterogeneity of the results that have been reported that are influenced by the limited testing conditions, sample size, and lack of publicly available datasets. Finally, there is potential to standardize the features computed through the time or frequency domains and evaluate and compare the diverse set of settings of time-frequency methods used for feature extraction and signal representation.
6 citations
References
More filters
Proceedings Article•
03 Dec 2012TL;DR: The state-of-the-art performance of CNNs was achieved by Deep Convolutional Neural Networks (DCNNs) as discussed by the authors, which consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax.
Abstract: We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.
73,978 citations
37,017 citations
"RoadCare: A Deep-learning Based App..." refers methods in this paper
...It first uses a watershed algorithm [4] to perform road segmentation, followed by Otsu’s binarization [39] and morphological thinning [5] to finally generate a topological skeleton of the pothole image....
[...]
Proceedings Article•
06 Jul 2015TL;DR: Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin.
Abstract: Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful parameter initialization, and makes it notoriously hard to train models with saturating nonlinearities. We refer to this phenomenon as internal covariate shift, and address the problem by normalizing layer inputs. Our method draws its strength from making normalization a part of the model architecture and performing the normalization for each training mini-batch. Batch Normalization allows us to use much higher learning rates and be less careful about initialization, and in some cases eliminates the need for Dropout. Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin. Using an ensemble of batch-normalized networks, we improve upon the best published result on ImageNet classification: reaching 4.82% top-5 test error, exceeding the accuracy of human raters.
30,843 citations
"RoadCare: A Deep-learning Based App..." refers methods in this paper
...For training the NNdeep architecture, batch normalization [21] was used at the end of the first and second convolutional layers 4....
[...]
TL;DR: A new graphical display is proposed for partitioning techniques, where each cluster is represented by a so-called silhouette, which is based on the comparison of its tightness and separation, and provides an evaluation of clustering validity.
14,144 citations
Additional excerpts
...ette [42] analysis on our Lunl dataset....
[...]
07 May 2006
TL;DR: A novel scale- and rotation-invariant interest point detector and descriptor, coined SURF (Speeded Up Robust Features), which approximates or even outperforms previously proposed schemes with respect to repeatability, distinctiveness, and robustness, yet can be computed and compared much faster.
Abstract: In this paper, we present a novel scale- and rotation-invariant interest point detector and descriptor, coined SURF (Speeded Up Robust Features). It approximates or even outperforms previously proposed schemes with respect to repeatability, distinctiveness, and robustness, yet can be computed and compared much faster.
This is achieved by relying on integral images for image convolutions; by building on the strengths of the leading existing detectors and descriptors (in casu, using a Hessian matrix-based measure for the detector, and a distribution-based descriptor); and by simplifying these methods to the essential. This leads to a combination of novel detection, description, and matching steps. The paper presents experimental results on a standard evaluation set, as well as on imagery obtained in the context of a real-life object recognition application. Both show SURF's strong performance.
13,011 citations