Video-based trajectory extraction with deep learning for High-Granularity Highway Simulation (HIGH-SIM)

Shi, Xiaowei; Zhao, Dongfang; Yao, Handong; Li, Xiaopeng; Hale, Douglas A.; Ghiasi, Amir

doi:10.1016/j.commtr.2021.100014

Cited by 42 publications

(12 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another related research problem is filling the missing vehicle trajectory data points. Shi et al proposed a Monte-Carlo-based lane marking identification approach [36] to extract the vehicle trajectory data. In [37], the authors proposed a car-following-based (CF-based) vehicle trajectory connection method that can fill missing data points caused by detection errors.…”

Section: Deep Learning-based Methodsmentioning

confidence: 99%

Attention-Driven Recurrent Imputation for Traffic Speed

Zhang

et al. 2022

IEEE Open J. Intell. Transp. Syst.

View full text Add to dashboard Cite

In practice, traffic data collection is often warned by missing data due to communication errors, sensor failures, storage loss, among other factors, leading to impaired data collection and hampering the effectiveness of downstream applications. However, existing imputation approaches focus exclusively on estimating the lost value from incomplete observations and ignore historical data. In this paper, we propose a novel neural network model, namely, Attention-Driven Recurrent Imputation Network (ADRIN), to address the problem of missing traffic data. Specifically, in ADRIN, we devise an Imputation-targeted Long Short-Term Memory (LSTM-I) module for filling in missing data. Meanwhile, we consider the periodicity of historical data and design a historical average calculation module in ADRIN. On this basis, we employ the multi-head self-attention mechanism for further extracting latent temporal features from the output of the two modules. ADRIN is capable of modeling both incomplete observation inputs and historical averages independently to estimate the missing values. We conducted comprehensive experiments on three real-world traffic datasets, to demonstrate that ADRIN consistently outperforms other baselines in a variety of scenarios. Furthermore, ablation experiments are conducted on the various modules of the model, and it is concluded that historical data can significantly enhance the imputation effect.

show abstract

Section: Deep Learning-based Methodsmentioning

confidence: 99%

Attention-Driven Recurrent Imputation for Traffic Speed

Zhang

et al. 2022

IEEE Open J. Intell. Transp. Syst.

View full text Add to dashboard Cite

show abstract

“…These advances, along with the increasing prevalence of aerial drones, have enabled recent research efforts to revisit the task of vehicle trajectory extraction and make marked advancements to the state of the art. The HighD, [2], ExiD [4], AUTOMATUM [5], and HIGH-SIM [6] datasets all utilize aerial imagery shot from either drone or helicopter-mounted cameras to produce complete highway vehicle trajectory data, and the Third Generation Simulation (TGSIM) [79] is a similar in-progress effort designed to capture trajectory data containing deployed automated vehicle technologies. Similarly, the pNEUMA [3], inD [80], rounD [81], OpenDD [82], Interaction [83] and CitySim [84] datasets utilize drones or swarms of drones to study complex urban vehicle and pedestrian interactions in more detail.…”

Section: Emerging Observation Technologiesmentioning

confidence: 99%

I-24 MOTION: An instrument for freeway traffic science

Gloudemans¹,

Wang²,

Ji³

et al. 2023

Preprint

View full text Add to dashboard Cite

The Interstate-24 MObility Technology Interstate Observation Network (I-24 MOTION) is a new instrument for traffic science located near Nashville, Tennessee. I-24 MOTION consists of 276 pole-mounted high-resolution traffic cameras that provide seamless coverage of approximately 4.2 miles I-24, a 4-5 lane (each direction) freeway with frequently observed congestion. The cameras are connected via fiber optic network to a compute facility where vehicle trajectories are extracted from the video imagery using computer vision techniques. Approximately 230 million vehicle miles of travel occur within I-24 MOTION annually. The main output of the instrument are vehicle trajectory datasets that contain the position of each vehicle on the freeway, as well as other supplementary information, vehicle dimensions, and class. This article describes the design and creation of the instrument, and provides the first publicly available datasets generated from the instrument. The datasets published with this article contains at least 4 hours of vehicle trajectory data for each of 10 days. As the system continues to mature, all trajectory data will be made publicly available at i24motion.org.

show abstract

“…From a review of the existing studies related to expressways and Interstate highways, the authors found that most of the current studies are based on the NGSIM dataset, which was collected by the U.S. Department of Transportation’s Joint Program Office for Intelligent Transportation Systems in 2005 using cameras on top of high-rise buildings ( 11 ). From NGSIM, the dataset recorded on U.S. Highway 101 (US-101) has 15 min of congestion data, covering a maximum of 640 m. It should be noted that the NGSIM US-101 dataset has the best data quality compared with the other three NGSIM datasets, and some new datasets are often compared with the quality of US-101 ( 13 ). In recent years, some institutions or researchers have published additional datasets, which improves the accuracy and length of trajectory data.…”

mentioning

confidence: 99%

“…For example, the highD dataset is bird’s-eye view imaging recorded by unmanned aerial vehicles (UAVs) along German expressways during 2017–2018, including a total of 60 records from six different locations, with an average duration of 17 min (16.5 h in total) and a spatial range of about 420 m ( 12 ). In addition, High-SIM is a vehicle trajectory dataset recorded on the 2,438 m length of Interstate highway 75 (I-75) in Florida, U.S.A., which contains data of three lanes and one off-ramp captured by three 8K cameras on the helicopter from 4:15 to 6:15 p.m. ( 13 ).…”

mentioning

confidence: 99%

MAGIC Dataset: Multiple Conditions Unmanned Aerial Vehicle Group-Based High-Fidelity Comprehensive Vehicle Trajectory Dataset

Zhong

Wang

et al. 2022

Transportation Research Record

View full text Add to dashboard Cite

High-fidelity vehicle trajectory data contain rich spatiotemporal characteristics and play a major role in the field of transportation research, for example, driving behavior models, traffic flow models, traffic state identification, and driver assistance strategies. There are some coverage issues with the existing datasets recorded in continuous traffic flow facilities, which may limit the further development of research on the safety and efficiency of continuous traffic flow facilities. For example, the observation areas of these datasets lack different traffic states and section types. Therefore, it is necessary to collect a new trajectory dataset. This paper proposes a detailed plan for aerial photography by unmanned aerial vehicle (UAV) group. An experiment was conducted from 7:40 to 10:40 a.m. on a section of the Shanghai Inner Ring, Shanghai, China, with a total length of 4,000 m in both directions including a large radius curve and six ramps. The trajectory dataset, named MAGIC, is extracted and compared with NGSIM US-101 and HIGH-SIM from the aspects of experiment field, traffic states, and so on. Further, to illustrate the advantages of the MAGIC dataset incorporating different traffic states and section types, the MAGIC dataset is evaluated from the three aspects of traffic congestion state, fundamental diagram, and traffic conflict. Overall, there are significant differences in macroscopic traffic flow and traffic safety characteristics between road sections with different section types or traffic states. Therefore, the proposed method presents some unique advantages and may perform effectively in many fields.

show abstract

Video-based trajectory extraction with deep learning for High-Granularity Highway Simulation (HIGH-SIM)

Cited by 42 publications

References 24 publications

Attention-Driven Recurrent Imputation for Traffic Speed

Attention-Driven Recurrent Imputation for Traffic Speed

I-24 MOTION: An instrument for freeway traffic science

MAGIC Dataset: Multiple Conditions Unmanned Aerial Vehicle Group-Based High-Fidelity Comprehensive Vehicle Trajectory Dataset

Contact Info

Product

Resources

About