Real-time vehicle detection under complex road conditions

With the rapid development of Internet of Things (IoT) and UAV technology, for the whole IoT system of vehicle detection, the middle and high level of information transmission and server processing has made a breakthrough, so at the bottom of the real-time detection of the vehicle by the UAV is the key to the whole system. However, UAV vehicle detection faces the challenges of too many small targets in the image leading to low detection accuracy, limited hardware platform resources requiring control of model size, and real-time detection requiring high inference speed. Aiming at the above problems, we propose a lightweight model PrFu-YOLO based on YOLOv8 improvement, which achieves a good balance between the accuracy, inference speed, and model size. And it realizes real-time vehicle detection embedded in an UAV platform. To solve the problem of low vehicle detection accuracy, we design a new structure PrFuFPN based on adding a small target detection layer to achieve more advanced feature fusion. To address the limited resources of the platform and the problem of real-time vehicle detection, we add GhostConv to the structure and constantly try to adjust the parameters of the network. Finally, extensive experiments were conducted on the VisDrone2019 and CARPK data sets to fully evaluate the model. Compared to YOLOv8s on the VisDrone2019 test set, mAP50 was improved by 10.05%, mA95 by 14.54%, the number of parameters was reduced by 8.13%, and the model size was reduced by 12.5%, while the FPS of 67.13 fully met the needs of real-time detection.

Increasing driving safety by virtue of advanced technology requires real-time and accurate detection of vehicles in far and close distance, which pose a threat to the host vehicle. This paper presents a real-time on-road vehicle detection algorithm based on monocular vision. First, auto-adapted threshold segmentation is proposed to extract shadow features. Then, a special mask is used for morphology computing to retain the features of vehicles in far and close distance. And, vanishing point constraint is applied for the fast verification of vehicles. Finally, the tracking of vehicles assists to stabilize the detection results. The experiments show that the average processing speed reaches 39 frames/ms, and the fast detection of vehicle under different weather conditions in the day time can also work accurately.

Vehicle detection in intelligent transportation systems (ITS) is a very important and challenging task in traffic monitoring. The difficulty of this task is to accurately locate and classify relatively small vehicles in complex scenes. To solve these problems, this paper proposes a modified one-stage detector based on background prediction and group normalization to realize real-time and accurate detection of traffic vehicles. The one-stage detector firstly adds a module to adjust the width and height of anchors and predict the target background, which avoids the problem of the target vehicle missing detection or wrong detection due to the influence of the complicated environments. Then, group normalization and the loss function based on weight attenuation can improve the one-stage detector performance in the training process. The experimental results on traffic monitoring datasets indicate that the improved one-stage detector is superior to the other neural network models in terms of precision at 95.78%.

This paper presents a robustness benchmark evaluation and optimization for vehicle detection. Real-time vehicle detection has become an essential means of data perception in the transportation field, covering various aspects such as intelligent transportation systems, video surveillance, and autonomous driving. However, evaluating and optimizing the robustness of vehicle detection in real traffic scenarios remains challenging. When data distributions change, such as the impact of adverse weather or sensor damages, model reliability cannot be guaranteed. We first conducted a large-scale robustness benchmark evaluation for vehicle detection. Analysis revealed that adverse weather, motion, and occlusion are the most detrimental factors to vehicle detection performance. The impact of color changes and noise, while present, is relatively less pronounced. Moreover, the robustness of vehicle detection is closely linked to its baseline performance and model size. And as the severity of corruption intensifies, the performance of models experiences a sharp drop. When the data distribution of images changes, the features of the vehicles that the model focuses on are weakened, making the activation level of the targets significantly reduced. By evaluation, we provided guidance and direction for optimizing detection robustness. Based on these findings, we propose TDIRM, a traffic-degraded image restoration model based on stable diffusion, designed to efficiently restore degraded images in real traffic scenarios and thereby enhance the robustness of vehicle detection. The model introduces an image semantics encoder (ISE) module to extract features that align with the latent description of the real background while excluding degradation-related information. Additionally, a triple control embedding attention (TCE) module is proposed to fully integrate all condition controls. Through a triple condition control mechanism, TDIRM achieves restoration results with high fidelity and consistency. Experimental results demonstrate that TDIRM improves vehicle detection mAP by 6.92% on real dense fog data, especially for small distant vehicles that were severely obscured by fog. By enabling semantic-structural-content collaborative optimization within the diffusion framework, TDIRM establishes a novel paradigm for traffic scene image restoration.

Vehicle detection and tracking is a significant part in auxiliary vehicle driving system. Using the traditional detection method based on image information has encountered enormous difficulties, especially in complex background. To solve this problem, a detection method based on deep learning, Faster R-CNN, which has very high detection accuracy and flexibility, is introduced. An algorithm of target tracking with the combination of Camshift and Kalman filter is proposed for vehicle tracking. The computation time of Faster R-CNN cannot achieve realtime detection. We use multi-thread technique to detect and track vehicle by parallel computation for real-time application.

Real-time vehicle detection using histograms of oriented gradients and AdaBoost classification

Deep learning-based classification and detection algorithms have emerged as a powerful tool for vehicle detection in intelligent transportation systems. The limitations of the number of high-quality labeled training samples makes the single vehicle detection methods incapable of accomplishing acceptable accuracy in road vehicle detection. This paper presents detection and classification of vehicles on publicly available datasets by utilizing the YOLO-v5 architecture. This paper’s findings utilize the concept of transfer learning through fine tuning the weights of the pre-trained YOLO-v5 architecture. To employ the concept of transfer learning, extensive data sets of images and videos of the congested traffic patterns were collected by the authors. These datasets were made more comprehensive by pointing various attributes, for instance high- and low-density traffic patterns, occlusions, and different weather circumstances. All of these gathered datasets were manually annotated. Ultimately, the improved YOLO-v5 structure becomes accustomed to any difficult traffic patterns. By fine-tuning the pre-trained network through our datasets, our proposed YOLO-v5 has exceeded several other traditional vehicle detection methods in terms of detection accuracy and execution time. Detailed simulations performed on the PKU, COCO, and DAWN datasets demonstrate the effectiveness of the proposed method in various challenging situations.

The Vehicle detection is important for the enhancement of transportation systems and which is efficient for traffic management, improved road safety and accurate data collection by automatically identifying and tracking vehicles on roads which enables features like traffic signal optimization, speed measurement and accident detection ultimately contributing to a smoother and safer driving experience for everyone. Here we have built a real-time project which can detect car, bus, motorcycle and truck on the basis of algorithm called YOLOV8 (You Only Look Once Version 8). It is a computer vision technique which is renowned for its real-time object detection capabilities, providing optimal balance between speed and accuracy. The model is trained using PyTorch and leverages Convolutional Neural Network (CNN) for feature extraction.

We propose a method of real-time moving vehicle detection using infrared thermal images. It can detect moving vehicles robustly even for bad environments compared with conventional vehicle detection methods using visible light cameras. It can also measure the size of each vehicle around the clock. The algorithm we propose for this detection is designed for a high-speed processing without complicated calculations and also designed for a real-time vehicle detection by using a general-purpose personal computer. Experimental results of our method by use of infrared thermal traffic images both at daytime with vehicular shadows and at nighttime in darkness show that the vehicle detection accuracy is 100%.

Vehicle detection from short-range aerial image faces challenges including vehicle blocking, irrelevant object interference, motion blurring, color variation etc., leading to the difficulty to achieve high detection accuracy and real-time detection speed. In this paper, benefiting from the recent development in MobileNet family network engineering, we propose a compressed MobileNet which is not only internally resistant to the above listed challenges but also gains the best detection accuracy/speed tradeoff when comparing with the original MobileNet. In a nutshell, we reduce the bottleneck architecture number during the feature map downsampling stage but add more bottlenecks during the feature map plateau stage, neither extra FLOPs nor parameters are thus involved but reduced inference time and better accuracy are expected. We conduct experiment on our collected 5-k short-range aerial images, containing six vehicle categories: truck, car, bus, bicycle, motorcycle, crowded bicycles and crowded motorcycles. Our proposed compressed MobileNet achieves 110 FPS (GPU), 31 FPS (CPU) and 15 FPS (mobile phone), 1.2 times faster and 2% more accurate (mAP) than the original MobileNet.

In the field of intelligent transportation, accurate vehicle detection, tracking, and re-identification are essential tasks that enable real-time monitoring, congestion management, and safety improvements. To address these needs in high-traffic highway environments, this study proposes a multi stage traffic flow model combining deep learning and metric learning. The model leverages the Segment Anything Model for vehicle detection, utilizing language-prompting to automate segmentation, thereby reducing manual adjustments and improving adaptability across complex traffic scenarios. For vehicle tracking, the model employs the StrongSORT algorithm, integrated with mask-based tracking to enhance recognition coherence and maintain resilience against occlusions, especially in congested conditions. Additionally, the PP-OCR module accurately extracts timestamps to support interval speed measurement across multiple viewpoints, using vehicle re-identification results for precise multi-camera tracking. Experimental results demonstrate that the proposed model significantly outperforms state-of-the-art methods, achieving higher mean average precision and tracking accuracy in both high-density and challenging traffic conditions, highlighting its robustness and suitability for intelligent transportation systems.

With the increase in the number of vehicles in our country, traffic accidents have become frequent. The real-time detection of dense traffic vehicles is particularly important, which can promote the development of autonomous driving and lay a good foundation for intelligent transportation. Aiming at the problems such as high vehicle overlap rate, high small target missing rate and poor real-time detection in traffic vehicle data set, a traffic vehicle target detection method based on improved YOLOv7 was proposed. Firstly, the neck network structure of the original YOLOv7 is lightweight and introduced into the GhostSlimPAN paradigm structure to reduce the number of model parameters and operation cost, and improve the accuracy of detection of small target vehicles. Then, the original loss function CIoU is improved to replusion loss function to reduce the missed and false detection rate caused by mutual occlusion between vehicles. In order to verify the effect of the improved model, the vehicle dataset was selected for testing and verification. The experimental results show that compared with the original YOLOv7 algorithm, the [email protected] of the improved YOLOv7 algorithm on this dataset reaches 88.2%, which is 1.8% higher than that of the benchmark network. The improved algorithm can be applied to vehicle target detection in daily environment, providing key technical support for intelligent transportation and autonomous driving.

This paper introduces YOLOv11-SEG, a real-time vehicle detection with instance segmentation using deep learning for Advanced Driver Assistance Systems (ADAS) and autonomous driving. The C3k2 modules for more accurate feature extraction and C2PSA attention for the treatment of small or occluded objects are added to the YOLO architecture. Trained with a fiveclass vehicle dataset of 6,500 instances, the model achieves box detection precision of 94.5% and segmentation precision of 94.9%, with recall of 94.4% (box) and 93.3% (mask) showing its high sensitivity. mAP@0.5 values of 97.4% (detection) and 97.1% (segmentation) show accurate object localization, while mAP@0.5-0.95 values of 90.4% and 88.5% show robust performance at varying IoU thresholds. Qualitative tests demonstrate consistency in performance in unfavorable conditions of fog, low light, and occlusion, showing the suitability of YOLOv11-SEG for real-world deployment in ADAS. Future work will be focused on accelerating the speed of inference and reducing the computational complexity, enhancing the applicability of the model for safe, efficient autonomous driving.

We developed a low-cost, high-precision monocular visual ranging system based on the You Only Look Once version 5 (YOLOv5) object detection algorithm, addressing the limitations of traditional methods by integrating deep learning with computer vision.The YOLOv5 architecture is optimized through the adoption of the cross stage partial darknet backbone network and the Mosaic data augmentation strategy.The architecture improves vehicle detection accuracy with a mean average precision at intersection over union threshold 0.5 of 0.5719, and the inference speed of 140 frames per second.Multiscale object detection capabilities are further enhanced using an adaptive anchor box generation mechanism.The monocular ranging model was constructed by combining camera calibration parameters with perspective geometry principles.Bounding box information produced by YOLOv5 is mapped to three-dimensional spatial distances through the relationship between object pixel dimensions and actual physical dimensions, effectively mitigating the scale ambiguity inherent in monocular vision.The experimental results showed that the system maintained average ranging errors within engineering-acceptable limits under diverse lighting and weather conditions while satisfying real-time requirements.Compared with traditional geometric ranging benchmarks, the optimized model showed a 12% improvement in mean absolute error and root mean square error with an overall accuracy enhancement of 16.88%.The results confirm that the developed monocular visual ranging model balances cost and performance and is a reliable method for intelligent transportation systems.By functioning as a robust software-defined virtual sensor, the system offers significant engineering application value and contributes to the advancement of intelligent driving technologies.

This paper presents a lightweight and efficient vehicle detection and tracking method for smart intersection environments using a low-cost, low-resolutionFMCW(Frequency Modulated Continuous Wave) LiDAR(Light Detecting And Ranging) sensor.We propose a framework that leverages Doppler-based velocity information fromFMCWLiDAR to facilitate point cloud accumulation. This leads to more effective background removal and robust multi-object tracking, even with low-resolution LiDAR data. The point cloud accumulation technique compensates for the sparsity of points representing vehicles by leveraging the radial velocity of detected points, resulting in denser and more accurate representations of vehicles. Background removal and lane segmentation are applied to effectively remove static objects and improve detection accuracy. The Kalman filter incorporates both position and velocity information of detected objects, enabling reliable tracking even when frame rates are low or objects move rapidly. The overall system is designed for real-time operation on edge devices, making it suitable for deployment in practical smart city scenarios. Experimental validation using real-world road data demonstrates that the proposed method significantly enhances detection and tracking performance compared to conventional approaches.