A new scheme of vehicle detection for severe weather based on multi-sensor fusion

Wang, Zhangu; Zhan, Jianming; Li, Ye; Zhong, Zhaohui; Cao, Zikun

doi:10.1016/j.measurement.2022.110737

Cited by 21 publications

(5 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Sensor fusion is recommended to increase the detection performance of a framework where the data from the camera can be merged with radar, light detection and ranging (LiDAR), etc., to obtain better detection results. The authors present a multi-sensor fusion approach to improve vehicle detection in bad weather [ 23 ]. A LightGBM model is used for vehicle target extraction using radar.…”

Section: Related Workmentioning

confidence: 99%

Analyzing Performance of YOLOx for Detecting Vehicles in Bad Weather Conditions

Ashraf,

Hur,

Kim

et al. 2024

Sensors

View full text Add to dashboard Cite

Recent advancements in computer vision technology, developments in sensors and sensor-collecting approaches, and the use of deep and transfer learning approaches have excelled in the development of autonomous vehicles. On-road vehicle detection has become a task of significant importance, especially due to exponentially increasing research on autonomous vehicles during the past few years. With high-end computing resources, a large number of deep learning models have been trained and tested for on-road vehicle detection recently. Vehicle detection may become a challenging process especially due to varying light and weather conditions like night, snow, sand, rain, foggy conditions, etc. In addition, vehicle detection should be fast enough to work in real time. This study investigates the use of the recent YOLO version, YOLOx, to detect vehicles in bad weather conditions including rain, fog, snow, and sandstorms. The model is tested on the publicly available benchmark dataset DAWN containing images containing four bad weather conditions, different illuminations, background, and number of vehicles in a frame. The efficacy of the model is evaluated in terms of precision, recall, and mAP. The results exhibit the better performance of YOLOx-s over YOLOx-m and YOLOx-l variants. YOLOx-s has 0.8983 and 0.8656 mAP for snow and sandstorms, respectively, while its mAP for rain and fog is 0.9509 and 0.9524, respectively. The performance of models is better for snow and foggy weather than rainy weather sandstorms. Further experiments indicate that enhancing image quality using multiscale retinex improves YOLOx performance.

show abstract

Section: Related Workmentioning

confidence: 99%

Analyzing Performance of YOLOx for Detecting Vehicles in Bad Weather Conditions

Ashraf,

Hur,

Kim

et al. 2024

Sensors

View full text Add to dashboard Cite

show abstract

“…The method combines the advantages of millimeter-wave radar and other detectors and also improves the accuracy of the traffic detection system. In [ 29 ], millimeter-wave radar and infrared cameras were selected as sensors for vehicle detection. The detection information of millimeter-wave radar can extract the vehicle region of interest (ROI) from infrared images, and infrared images can compensate for the low resolution of the radar.…”

Section: Related Workmentioning

confidence: 99%

A Trajectory Compensation Method Considering the Car-Following Behavior for Data Missing of Millimeter-Wave Radar in Roadside Detection Applications

Zhang

Liu

Teng

2023

Sensors

View full text Add to dashboard Cite

Concerning roadside traffic detection applications, and to address the millimeter-wave radar’s missing data problem caused by target occlusion or the absence of features in low-speed conditions, this paper proposes a trajectory compensation method regarding car-following behavior. Referring to the installation scheme of the detector, a coordinate transformation method is presented to unify the radar spatial coordinates with the road coordinates. Considering the driver’s car-following behavior, the optimal velocity model (OV), full velocity difference model (FVD), and the full velocity difference and acceleration (FVDA) model are applied for tracking the vehicle’s trajectory related to the movement of the vehicle ahead. Finally, a data compensation scheme is presented. Taking actual trajectory data as samples, the proposed methods are verifiably useful for compensating for missing data and reconstructing target trajectories. Statistical results of different missing data trajectories demonstrate the rationality of the application of car-following models for the missing data compensation, and the FVDA model performs well compared with the OV and FVD models.

show abstract

“…In addition, recognising and tracking cars have been difficult problems for conventional computer vision and image processing research because to obstacles such as partial or complete occlusion of objects, camera shake, variable image quality, and weather conditions such as rain, snow, and wind [17]. These obstacles make it difficult to reliably identify and track cars, and in certain circumstances such systems may not function at all [18].…”

Section: Introductionmentioning

confidence: 99%

Yus - A Deep Learning Algorithm for Collision Avoidance through Object and Vehicle Detection

Beg¹,

Ismail²,

Miah³

et al. 2023

ARASET

View full text Add to dashboard Cite

One of the safety features that can alert drivers to the presence of other vehicles and reduce the risk of collisions is vehicle detection. In this study, the objective was to create a system for detecting vehicles, motorcycles, and traffic signals on the roads in University Malaysia Pahang using object detection techniques. The video was taken through Go-Pro camera to capture video footage of traffic objects on the roads in the district, which was then analysed using the YOLO-V8 deep learning algorithm. The system was trained on a pre-existing dataset of 1,068 images, with 70% of the dataset used for training and 30% for testing. After conducting a performance validation, the system achieved a mean average precision of 88.2% on training dataset and was able to detect different types of vehicles such as cars, motorcycles, and traffic lights. The results of this study could be beneficial for road safety authorities and researchers interested in developing intelligent transportation systems.

show abstract

A new scheme of vehicle detection for severe weather based on multi-sensor fusion

Cited by 21 publications

References 47 publications

Analyzing Performance of YOLOx for Detecting Vehicles in Bad Weather Conditions

Analyzing Performance of YOLOx for Detecting Vehicles in Bad Weather Conditions

A Trajectory Compensation Method Considering the Car-Following Behavior for Data Missing of Millimeter-Wave Radar in Roadside Detection Applications

Yus - A Deep Learning Algorithm for Collision Avoidance through Object and Vehicle Detection

Contact Info

Product

Resources

About