Traffic Flow Statistics Research Articles

Traffic flow detection method based on improved SSD algorithm for intelligent transportation system.

With the development of the new generation communication system in China, the application of intelligent transportation system is more extensive, which brings higher demands for vehicle flow detection and monitoring. Traditional traffic flow detection modes often cannot meet the high statistical accuracy requirement and high-speed detection simultaneously. Therefore, an improved Inception module is integrated into the single shot multi box detector algorithm. An intelligent vehicle flow detection model is constructed based on the improved single shot multi box detector algorithm. According to the findings, the convergence speed of the improved algorithm was the fastest. When the test sample was the entire test set, the accuracy and precision values of the improved method were 93.6% and 96.0%, respectively, which were higher than all comparison target detection algorithms. The experimental results of traffic flow statistics showed that the model had the highest statistical accuracy, which converged during the training phase. During the testing phase, except for manual statistics, all methods had the lowest statistical accuracy on motorcycles. The average accuracy and precision of the designed model for various types of images were 96.9% and 96.8%, respectively. The calculation speed of this intelligent model was not significantly improved compared to the other two intelligent models, but it was significantly higher than manual monitoring methods. Two experimental data demonstrate that the intelligent vehicle flow detection model designed in this study has higher detection accuracy. The calculation speed has no significant difference compared with the traditional method, which is helpful to the traffic flow management in intelligent transportation system.

Real-Time Traffic Flow Statistics Based on Dual-Granularity Classification

Article Real-Time Traffic Flow Statistics Based on Dual-Granularity Classification Yanchao Bi, Yuyan Yin, Xinfeng Liu *, Xiushan Nie, Chenxi Zou, and Junbiao Du 1 School of Computer Science and Technology, Shandong Jianzhu University, Jinan 250101, China * Correspondence: liuxinfeng18@sdjzu.edu.cn Received: 17 June 2023 Accepted: 13 September 2023 Published: 26 September 2023 Abstract: Traffic detection devices can cause accuracy degradation over time. Considering problems such as time-consuming and laborious manual statistics, high misdetection probabilities, and model tracking failures, there is an urgent need to develop a deep learning model (which can stably achieve detection accuracy over 90%) to evaluate whether the device accuracy still satisfies the requirements or not. In this study, based on dual-granularity classification, a real-time traffic flow statistics method is proposed to address the above problems. The method is divided into two stages. The first stage uses YOLOv5 to acquire all the motorized and non-motorized vehicles appearing in the scene. The second stage uses EfficientNet to acquire the motorized vehicles obtained in the previous stage and classify such vehicles into six categories. Through this dual-granularity classification, the considered problem is simplified and the probability of false detection is reduced significantly. To correlate the front and back frames of the video, vehicle tracking is implemented using DeepSORT, and vehicle re-identification is implemented in conjunction with the ResNet50 model to improve the tracking accuracy. The experimental results show that the method used in this study solves the problems of misdetection and tracking effectively. Moreover, the proposed method achieves 98.7% real-time statistical accuracy by combining the two-lane counting method.

Recognition new energy vehicles based on improved YOLOv5.

In the field of Intelligent Traffic Systems (ITS), vehicle recognition is a hot research topic. Although different types of vehicles can already be recognized, further identification and statistics of new energy and fuel vehicles in unknown and complex environments remain a challenging task. In this paper, we propose a New Energy Vehicle Recognition and Traffic Flow Statistics (NEVTS) approach. Specifically, we first utilized the You Only Look Once v5 (YOLOv5) algorithm to detect vehicles in the target area, in which we applied Task-Specific Context Decoupling (TSCODE) to decouple the prediction and classification tasks of YOLOv5. This approach significantly enhanced the performance of vehicle detection. Then, track them upon detection. Finally, we use the YOLOv5 algorithm to locate and classify the color of license plates. Green license plates indicate new energy vehicles, while non-green license plates indicate fuel vehicles, which can accurately and efficiently calculate the number of new energy vehicles. The effectiveness of the proposed NEVTS in recognizing new energy vehicles and traffic flow statistics is demonstrated by experimental results. Not only can NEVTS be applied to the recognition of new energy vehicles and traffic flow statistics, but it can also be further employed for traffic timing pattern extraction and traffic situation monitoring and management.

HB-YOLO: An Improved YOLOv7 Algorithm for Dim-Object Tracking in Satellite Remote Sensing Videos

The You Only Look Once (YOLO) series has been widely adopted across various domains. With the increasing prevalence of continuous satellite observation, the resulting video streams can be subjected to intelligent analysis for various applications, such as traffic flow statistics, military operations, and other fields. Nevertheless, the signal-to-noise ratio of objects in satellite videos is considerably low, and their size is often smaller, ranging from tens to one percent, when compared to those taken by drones and other equipment. Consequently, the original YOLO algorithm’s performance is inadequate when detecting tiny objects in satellite videos. Hence, we propose an improved framework, named HB-YOLO. To enable the backbone to extract features, we replaced the universal convolution with an improved HorNet that enables higher-order spatial interactions. We replaced all Extended Efficient Layer Aggregation Networks (ELANs) with the BoTNet attention mechanism to make the features fully fused. In addition, anchors were re-adjusted, and image segmentation was integrated to achieve detection results, which are tracked using the BoT-SORT algorithm. The experimental results indicate that the original algorithm failed to learn using the satellite video dataset, whereas our proposed approach yielded improved recall and precision. Specifically, the F1-score and mean average precision increased to 0.58 and 0.53, respectively, and the object-tracking performance was enhanced by incorporating the image segmentation method.

Residual Neural Networks for Origin–Destination Trip Matrix Estimation from Traffic Sensor Information

Traffic management and control applications require comprehensive knowledge of traffic flow data. Typically, such information is gathered using traffic sensors, which have two basic challenges: First, it is impractical or impossible to install sensors on every arc in a network. Second, sensors do not provide direct information on origin-to-destination (O–D) demand flows. Consequently, it is essential to identify the optimal locations for deploying traffic sensors and then enhance the knowledge gained from this link flow sample to forecast the network’s traffic flow. This article presents residual neural networks—a very deep set of neural networks—to the problem for the first time. The suggested architecture reliably predicts the whole network’s O–D flows utilizing link flows, hence inverting the standard traffic assignment problem. It deduces a relevant correlation between traffic flow statistics and network topology from traffic flow characteristics. To train the proposed deep learning architecture, random synthetic flow data was generated from the historical demand data of the network. A large-scale network was used to test and confirm the model’s performance. Then, the Sioux Falls network was used to compare the results with the literature. The robustness of applying the proposed framework to this particular combined traffic flow problem was determined by maintaining superior prediction accuracy over the literature with a moderate number of traffic sensors.

A novel traffic optimization method using GRU based deep neural network for the IoV system.

At present, China is moving towards the direction of "Industry 4.0". The development of the automobile industry, especially intelligent automobiles, is in full swing, which brings great convenience to people's life and travel. However, at the same time, urban traffic pressure is also increasingly prominent, and the situation of traffic congestion and traffic safety is not optimistic. In this context, the Internet of Vehicles (also known as "IoV") opens up a new way to relieve urban traffic pressure. Therefore, in order to further optimize the road network traffic conditions in the IoV environment, this research focuses on the traffic flow prediction algorithm on the basis of deep learning to enhance traffic efficiency and safety. First, the study investigates the short-time traffic flow prediction by combining the characteristics of the IoV environment. To address the issues that existing algorithms cannot automatically extract data features and the model expression capability is weak, the study chooses to build a deep neural network using GRU model in deep learning for short-time traffic flow prediction, thereby improving the prediction accuracy of algorithm. Secondly, a fine-grained traffic flow statistics approach suitable for the IoV situation is suggested in accordance with the deep learning model that was built. The algorithm sends the vehicle characteristic data obtained through GRU model training into the fine-grained traffic flow statistics algorithm, so as to realize the statistics of traffic information of various types of vehicles. The advantage of this algorithm is that it can well count the traffic flow of multiple lanes, so as to better predict the current traffic status and achieve traffic optimization. Finally, the IoV environment is constructed to confirm the effectiveness of the prediction model. The prediction results prove that the new algorithm has good performance in traffic flow statistics in different scenarios.

SAFE: EFFICIENT DDOS ATTACK DEFENSE WITH ELASTIC TRAFFIC FLOW INSPECTION IN SDN-BASED DATA CENTERS

In this paper, we propose an efficient distributed denial-of-Service (DDoS) Attack deFEnse solution, namely SAFE, which utilizes an elastic traffic flow inspection mechanism, for Software-Defined Networking (SDN) based data centers. In particular, we first examine a leaf-spine SDN-based data center network, which is highly vulnerable to volumetric DDoS attacks. Next, we develop a rank-based anomaly detection algorithm to recognize anomalies in the amount of incoming traffic. Then, for the traffic flow inspection, we introduce a component called DFI (Deep Flow Inspection) running an Open vSwitch (OvS) that can be dynamically initiated (as a virtual machine) on-demand to collect traffic flow statistics. By utilizing deep reinforcement learning-based traffic monitoring from our previous study, the DFIs can be protected from the flow-table overflow problem while providing more detailed traffic flow information. Afterward, a machine learning-based attack detector analyzes the gathered flow rule statistics to identify the attack, and appropriate policies are implemented if an attack is recognized. The experiment results show that the SAFE can effectively defend against volumetric DDoS attacks while assuring a reliable Quality-of-Service level for benign traffic flows in SDN-based data center networks. Specifically, for TCP SYN and UDP floods, the SAFE attack detection performance is improved by approximately 40% and 30%, respectively, compared to the existing SATA solution. Furthermore, the attack mitigation performance, the ratio of dropped malicious packets obtained by the SAFE is superior by approximately 48% (for TCP SYN flood) and 52% (for UDP flood) to the SATA.

FPGA-Based Vehicle Detection and Tracking Accelerator.

A convolutional neural network-based multiobject detection and tracking algorithm can be applied to vehicle detection and traffic flow statistics, thus enabling smart transportation. Aiming at the problems of the high computational complexity of multiobject detection and tracking algorithms, a large number of model parameters, and difficulty in achieving high throughput with a low power consumption in edge devices, we design and implement a low-power, low-latency, high-precision, and configurable vehicle detector based on a field programmable gate array (FPGA) with YOLOv3 (You-Only-Look-Once-version3), YOLOv3-tiny CNNs (Convolutional Neural Networks), and the Deepsort algorithm. First, we use a dynamic threshold structured pruning method based on a scaling factor to significantly compress the detection model size on the premise that the accuracy does not decrease. Second, a dynamic 16-bit fixed-point quantization algorithm is used to quantify the network parameters to reduce the memory occupation of the network model. Furthermore, we generate a reidentification (RE-ID) dataset from the UA-DETRAC dataset and train the appearance feature extraction network on the Deepsort algorithm to improve the vehicles' tracking performance. Finally, we implement hardware optimization techniques such as memory interlayer multiplexing, parameter rearrangement, ping-pong buffering, multichannel transfer, pipelining, Im2col+GEMM, and Winograd algorithms to improve resource utilization and computational efficiency. The experimental results demonstrate that the compressed YOLOv3 and YOLOv3-tiny network models decrease in size by 85.7% and 98.2%, respectively. The dual-module parallel acceleration meets the demand of the 6-way parallel video stream vehicle detection with the peak throughput at 168.72 fps.

Vehicular Flow Characterization: An Internet of Video Things-Based Solution

Intelligent transportation systems (ITS) have emerged as the optimal solution to address urban mobility challenges. However, to effectively implement ITS, detailed traffic flow statistics are imperative. Various solutions have been proposed, including intrusive/non-intrusive sensors and compute vision-based solutions. However, these solutions have limitations in the number of measured traffic flow parameters, cost or performance under different traffic conditions. To overcome these limitations, we propose an Internet-of-Video-Things (IoVT) based solution. The sensor node (fabricated using Raspberry Pi Zero W, Pi camera, power bank, and Wi-Fi device) can live-streaming roadside traffic video to a remote Dell server located at our lab with Camlytics (commercially available traffic analysis software) installed. The proposed solution was field tested with a 45-minute live-streamed video of 720p at 25 frames per second. Results show that the proposed solution can measure more traffic flow parameters than intrusive and non-intrusive sensors, with an accuracy of 84.3% for vehicle count and speed estimation. Other parameters were also calculated, such as time/distance headway, spatial/temporal densities, heat maps, and trajectories. Additionally, the proposed solution can count pedestrians with an accuracy of 76.3%.

Aerial Traffic Statistics Based on YOLOv5+DeepSORT

Traffic flow statistics, as an important part of intelligent transportation system, usually requires manual statistics, which is time-consuming and labor-intensive. In order to save manual labor and improve the statistical efficiency, this paper is based on the strategy of YOLOv5+DeepSORT to count the aerial traffic flow by UAV, and the results show that the statistical accuracy of this method is close to that of manual statistics, which has high practical value.

Estimating Vehicle and Pedestrian Activity from Town and City Traffic Cameras.

Traffic cameras are a widely available source of open data that offer tremendous value to public authorities by providing real-time statistics to understand and monitor the activity levels of local populations and their responses to policy interventions such as those seen during the COrona VIrus Disease 2019 (COVID-19) pandemic. This paper presents an end-to-end solution based on the Google Cloud Platform with scalable processing capability to deal with large volumes of traffic camera data across the UK in a cost-efficient manner. It describes a deep learning pipeline to detect pedestrians and vehicles and to generate mobility statistics from these. It includes novel methods for data cleaning and post-processing using a Structure SImilarity Measure (SSIM)-based static mask that improves reliability and accuracy in classifying people and vehicles from traffic camera images. The solution resulted in statistics describing trends in the ‘busyness’ of various towns and cities in the UK. We validated time series against Automatic Number Plate Recognition (ANPR) cameras across North East England, showing a close correlation between our statistical output and the ANPR source. Trends were also favorably compared against traffic flow statistics from the UK’s Department of Transport. The results of this work have been adopted as an experimental faster indicator of the impact of COVID-19 on the UK economy and society by the Office for National Statistics (ONS).

Machine Learning-based Distributed Denial of Service Attacks Detection Technique using New Features in Software-defined Networks

Software-Defined Networking is a new network architecture that separates control and data planes. It has central network control and programmability facilities, so it improves manageability, scaling, and performance. However, it may suffer from creating a single point of failure against the controller, which represents the network control plane. So, defending the controller against attacks such as a distributed denial of service attack is a valuable and urgent issue. The advances of this paper are to implement an accurate and significant method to detect this attack with high accuracy using machine learning-based algorithms exploiting new advanced features obtained from traffic flow information and statistics. The developed model is trained with kernel radial basis function. The technique uses advanced features such as unknown destination addresses, packets inter-arrival time, transport layer protocol header, and type of service header. To the best knowledge of the authors, the proposed approach of the paper had not been used before. The proposed work begins with generating both normal and attack traffic flow packets through the network. When packets reach the controller, it extracts their headers and performs necessary flow calculations to get the needed features. The features are used to create a dataset that is used as an input to linear support vector machine classifier. The classifier is used to train the model with kernel radial basis function. Methods such as Naive Bayes, K-Nearest Neighbor, Decision Tree, and Random Forest are also utilized and compared with the SVM model to improve the detection operation. Hence, suspicious senders are blocked and their information is stored. The experimental results prove that the proposed technique detects the attack with high accuracy and low false alarm, compared to other related techniques.

Traffic Flow Statistics Method and Practice Based on Roadside Video with YOLO_V3

Temporary traffic flow observation plays a pivotal role in transportation planning, consultation, and decision-making, and it supplements fixed long-term traffic observation data. Although substantial automated technology has been used for traffic observation, such technology is limited by the temporariness and uncertainty of observation points; moreover, these methods are difficult to use in temporary observations. In this study, the YOLO_V3 algorithm, which is based on deep learning, is used for vehicle detection based on roadside videos of temporary observations. Moreover, a secondary detection framework based on vehicle detection and traffic counting regions is proposed. Then, a traffic counting pattern with Kalman filter, Hungarian allocation, and perspective projection transformation is established. In addition, by collecting multiple sets of actual video data, the effectiveness of the method under different conditions is analyzed in terms of three indicators: camera intersection angle with the road, erection height, and road traffic density. Results show that at a camera height of 3 m and a roadside angle of 30°, the accuracy is approximately 95%. However, traffic flow accuracy drops to approximately 90% when vehicles are blocked by large buses and trucks during detection. In this study, the algorithm is tested for execution efficiency using 1080p video streams on a Windows 10 x 64 operating system, a 2080Ti graphics card, a 64 GB RAM, and an i7-7820X CPU. Results show that camera installation angle and height have no considerable effect on operation efficiency. Under low-density traffic, the frames per second (FPS) value is approximately 44; under high-density traffic, the FPS value drops to approximately 33, indicating that the method still has high execution efficiency and can be used for real-time video traffic counting.

Traffic Flow Statistics Method Based on Deep Learning and Multi-Feature Fusion

Traffic Flow Statistics Method Based on Deep Learning and Multi-Feature Fusion

Local adaptive road control crossroads

Exploring the properties of traffic flow, collecting and processing statistics on the marker indicators and criteria that characterize these properties is an important task on the road to the implementation of adaptive and intelligent traffic management systems in cities. The article provides an overview of the methods and indicators that allow forming a theoretical basis on the patterns of movement of transport on sections of the street and road network. This base is the source of the initial information for the introduction of adaptive control systems at isolated crossroads in urban environments.

Verifiable and Privacy-Preserving Traffic Flow Statistics for Advanced Traffic Management Systems

Crowdsourcing-based traffic monitoring plays an important role in advanced traffic management systems due to its high accuracy and low costs, but it may expose drivers real identities and sensitive locations that results in the privacy leakage of drivers. In this paper, we propose a crowdsourcing-based traffic monitoring scheme that enables a transportation management center (TMC) to achieve traffic flow statistics at road intersections in an efficient, verifiable, and privacy-preserving manner. Specifically, by integrating a homomorphic encryption primitive and a super-increasing sequence, traffic flow can be flexibly structured and encrypted by drivers, i.e., each drivers travel direction at T-junctions or crossroads is protected. As a middle-ware between drivers and TMC, roadside units (RSUs) are introduced to aggregate and further perturb the aggregated encrypted traffic flow based on a differential privacy mechanism. In this way, TMC is capable of acquiring the traffic flow statistics by decrypting the perturbed encrypted traffic flow, without disclosing each individual drivers traffic information. In addition, based on a lightweight commitment proof, the correctness of the encrypted drivers data can be guaranteed, i.e., a selfish driver cannot arbitrarily manipulate his data to poison the aggregated traffic flow. Finally, security analysis demonstrates that the proposed scheme satisfies all desirable security properties, including confidentiality, verifiability, unlinkability, and traceability. Extensive simulations are also conducted to show that the proposed scheme is efficient in terms of low computation and communication costs.

Optimized YOLOv3 Algorithm and Its Application in Traffic Flow Detections

In the intelligent traffic system, real-time and accurate detections of vehicles in images and video data are very important and challenging work. Especially in situations with complex scenes, different models, and high density, it is difficult to accurately locate and classify these vehicles during traffic flows. Therefore, we propose a single-stage deep neural network YOLOv3-DL, which is based on the Tensorflow framework to improve this problem. The network structure is optimized by introducing the idea of spatial pyramid pooling, then the loss function is redefined, and a weight regularization method is introduced, for that, the real-time detections and statistics of traffic flows can be implemented effectively. The optimization algorithm we use is the DL-CAR data set for end-to-end network training and experiments with data sets under different scenarios and weathers. The analyses of experimental data show that the optimized algorithm can improve the vehicles’ detection accuracy on the test set by 3.86%. Experiments on test sets in different environments have improved the detection accuracy rate by 4.53%, indicating that the algorithm has high robustness. At the same time, the detection accuracy and speed of the investigated algorithm are higher than other algorithms, indicating that the algorithm has higher detection performance.

Optimization of outdoor design temperature for summer ventilation for undersea road tunnel using field measurement and statistics

Based on the statistics of traffic flow in an urban undersea road tunnel, the temperature, humidity as well as the air velocity were tested. The impact of the temperature difference between the outdoor air temperature during evening peak traffic flow and the ventilation design temperature at 14:00 on ventilation rate was also analyzed in detail. For temperature control of undersea road tunnel, a new method of determining the outdoor design temperature for summer ventilation was proposed. The field measurement and statistical research show the followings: (1) There is a daily cyclical hump for the traffic flow, and the morning and evening traffic peaks are roughly distributed from 7:00 to 9:00 and from 16:00 to 18:00; (2) The air velocity fluctuates between 3.0 and 5.5 m/s; (3) There is a time misfit between the peak time of traffic flow and the time taken for outdoor design temperature for summer ventilation. When carrying out ventilation design of the undersea road tunnel, the outdoor air temperature 1 h after the start of evening traffic peak should be taken as the outdoor design temperature for summer ventilation in order to reduce ventilation rate and energy consumption by more than 10%.

Investigation of a Promoted You Only Look Once Algorithm and Its Application in Traffic Flow Monitoring

We propose a high-performance algorithm while using a promoted and modified form of the You Only Look Once (YOLO) model, which is based on the TensorFlow framework, to enhance the real-time monitoring of traffic-flow problems by an intelligent transportation system. Real-time detection and traffic-flow statistics were realized by adjusting the network structure, optimizing the loss function, and introducing weight regularization. This model, which we call YOLO-UA, was initialized based on the weight of a YOLO model pre-trained while using the VOC2007 data set. The UA-CAR data set with complex weather conditions was used for training, and better model parameters were selected through tests and subsequent adjustments. The experimental results showed that, for different weather scenarios, the accuracy of the YOLO-UA was ~22% greater than that of the YOLO model before optimization, and the recall rate increased by about 21%. On both cloudy and sunny days, the accuracy, precision, and recall rate of the YOLO-UA model were more than 94% above the floating rate, which suggested that the precision and recall rate achieved a good balance. When used for video testing, the YOLO-UA model yielded traffic statistics with an accuracy of up to 100%; the time to count the vehicles in each frame was less than 30 ms and it was highly robust in response to changes in scenario and weather.

GeoSOT-Based Spatiotemporal Index of Massive Trajectory Data

With the rapid development of global positioning technologies and the pervasiveness of intelligent mobile terminals, trajectory data have shown a sharp growth trend both in terms of data volume and coverage. In recent years, increasing numbers of LBS (location based service) applications have provided us with trajectory data services such as traffic flow statistics and user behavior pattern analyses. However, the storage and query efficiency of massive trajectory data are increasingly creating a bottleneck for these applications, especially for large-scale spatiotemporal query scenarios. To solve this problem, we propose a new spatiotemporal indexing method to improve the query efficiency of massive trajectory data. First, the method extends the GeoSOT spatial partitioning scheme to the time dimension and forms a global space–time subdivision scheme. Second, a novel multilevel spatiotemporal grid index, called the GeoSOT ST-index, was constructed to organize trajectory data hierarchically. Finally, a spatiotemporal range query processing method is proposed based on the index. We implement and evaluate the index in MongoDB. By comparing the range query efficiency and scalability of our index with those of the other two space–time composite indexes, we found that our approach improves query efficiency levels by approximately 40% and has better scalability under different data volumes.