Vehicle Detection Research Articles

Enhanced Object Detection in Autonomous Vehicles through LiDAR—Camera Sensor Fusion

To realize accurate environment perception, which is the technological key to enabling autonomous vehicles to interact with their external environments, it is primarily necessary to solve the issues of object detection and tracking in the vehicle-movement process. Multi-sensor fusion has become an essential process in efforts to overcome the shortcomings of individual sensor types and improve the efficiency and reliability of autonomous vehicles. This paper puts forward moving object detection and tracking methods based on LiDAR—camera fusion. Operating based on the calibration of the camera and LiDAR technology, this paper uses YOLO and PointPillars network models to perform object detection based on image and point cloud data. Then, a target box intersection-over-union (IoU) matching strategy, based on center-point distance probability and the improved Dempster–Shafer (D–S) theory, is used to perform class confidence fusion to obtain the final fusion detection result. In the process of moving object tracking, the DeepSORT algorithm is improved to address the issue of identity switching resulting from dynamic objects re-emerging after occlusion. An unscented Kalman filter is utilized to accurately predict the motion state of nonlinear objects, and object motion information is added to the IoU matching module to improve the matching accuracy in the data association process. Through self-collected data verification, the performances of fusion detection and tracking are judged to be significantly better than those of a single sensor. The evaluation indexes of the improved DeepSORT algorithm are 66% for MOTA and 79% for MOTP, which are, respectively, 10% and 5% higher than those of the original DeepSORT algorithm. The improved DeepSORT algorithm effectively solves the problem of tracking instability caused by the occlusion of moving objects.

A systematic study of traffic sign recognition and obstacle detection in autonomous vehicles

A systematic study of traffic sign recognition and obstacle detection in autonomous vehicles

Performance evaluation of YOLOv5 and YOLOv8 models in car detection

Vehicle detection stands out as a rapidly developing technology today and is further strengthened by deep learning algorithms. This technology is critical in traffic management, automated driving systems, security, urban planning, environmental impacts, transportation, and emergency response applications. Vehicle detection, which is used in many application areas such as monitoring traffic flow, assessing density, increasing security, and vehicle detection in automatic driving systems, makes an effective contribution to a wide range of areas, from urban planning to security measures. Moreover, the integration of this technology represents an important step for the development of smart cities and sustainable urban life. Deep learning models, especially algorithms such as You Only Look Once version 5 (YOLOv5) and You Only Look Once version 8 (YOLOv8), show effective vehicle detection results with satellite image data. According to the comparisons, the precision and recall values of the YOLOv5 model are 1.63% and 2.49% higher, respectively, than the YOLOv8 model. The reason for this difference is that the YOLOv8 model makes more sensitive vehicle detection than the YOLOv5. In the comparison based on the F1 score, the F1 score of YOLOv5 was measured as 0.958, while the F1 score of YOLOv8 was measured as 0.938. Ignoring sensitivity amounts, the increase in F1 score of YOLOv8 compared to YOLOv5 was found to be 0.06%.

Optimizing Vehicle Detection and Tracking Efficiency Through YOLO-Based Multi-Objective Approach

In traffic management, a careful detection of cars plays and monitors a vital role in ensuring safety and efficiency. This paper proposes a way to enhance the discovery of vehicles by dividing roads into remote areas and closing. The newly developed approach aims to define a vehicle and calculate vehicles in both fields. Yolov5M network is easy to discover and localize vehicles, exceed the traditional speed tracking methods and enable microorganisms. The proposed form achieves a higher accuracy of 0.833, outperform the Yolov5 Standard Form of 0.67. It also demonstrates the performance enhancement in the results of summons and mean average precision by keeping the training parameters reduction trend the same. Technically speaking, the split of the roads surface is divided into sections using the latest retail techniques and the fine tuning of the Yolov5M network to positively affect the detection and classification of vehicles.

Advanced Road Safety: Object Detection in Autonomous Vehicles using Deep Learning and IoT

Advanced Road Safety: Object Detection in Autonomous Vehicles using Deep Learning and IoT

The Comparative Study Between the Grid Size of Image Frame Using Image Subtraction and Pixel Expansion Cue for Free Obstacle Region Detection

This research investigates optimizing grid sizes for obstacle detection in Unmanned Aerial Vehicles (UAVs) using vision-based approaches. There were many proper computer visions algorithm that can be used to detect free obstacle region. One of it is by using grid-based approach. By dividing the image that taken by camera, the algorithm can choose the best path to go through.This study willexplore various grid sizes, from 3x4 to 24x32, to determine their effectiveness in detecting free obstacle regions in different environments. The methodology combines image subtraction, pixel expansion cues, and k-means segmentation, with images resized to 640x360 pixels and processed to display only two colors for simplification. The experiments, conducted using an iPhone 12 camera and analyzed with Spyder software and OpenCV, reveal that larger grids are more accurate in complex environments, albeit with higher computational time, whereas smaller grids are more efficient in simpler settings. The research identifies a grid size range of 12x16 to 16x24 as a balanced solution for various scenarios, highlighting the need for adaptive sizing and sophisticated algorithms in UAV obstacle detection.

An Enhanced Model for Detecting and Classifying Emergency Vehicles Using a Generative Adversarial Network (GAN)

The rise in autonomous vehicles further impacts road networks and driving conditions over the road networks. Cameras and sensors allow these vehicles to gather the characteristics of their surrounding traffic. One crucial factor in this environment is the appearance of emergency vehicles, which require special rules and priorities. Machine learning and deep learning techniques are used to develop intelligent models for detecting emergency vehicles from images. Vehicles use this model to analyze regularly captured road environment photos, requiring swift actions for safety on road networks. In this work, we mainly developed a Generative Adversarial Network (GAN) model that generates new emergency vehicles. This is to introduce a comprehensive expanded dataset that assists emergency vehicles detection and classification processes. Then, using Convolutional Neural Networks (CNNs), we constructed a vehicle detection model demonstrating satisfactory performance in identifying emergency vehicles. The detection model yielded an accuracy of 90.9% using the newly generated dataset. To ensure the reliability of the dataset, we employed 10-fold cross-validation, achieving accuracy exceeding 87%. Our work highlights the significance of accurate datasets in developing intelligent models for emergency vehicle detection. Finally, we validated the accuracy of our model using an external dataset. We compared our proposed model’s performance against four other online models, all evaluated using the same external dataset. Our proposed model achieved an accuracy of 85% on the external dataset.

A new network model for multiple object detection for autonomous vehicle detection in mining environment

AbstractConsidering the challenges of low multi‐object detection accuracy and difficulty in identifying small targets caused by challenging environmental conditions including irregular lighting patterns and ambient noise levels in the mining environment with autonomous electric locomotives. A new network model based on SOD−YOLOv5s−4L has been proposed to detect multi‐objects for autonomous electric locomotives in underground coal mines. Improvements have been applied in YOLOv5s to construct the SOD−YOLOv5s−4L model, by introducing the SIoU loss function to address the mismatch between real and predicted bounding box directions, facilitating the model to learn target position information more efficiently. This research introduces a decoupled head to enhance feature fusion and improve the positioning precision of the network model, enabling rapid capture of multi‐scale target features. Furthermore, the detection capability of the model has been increased by introducing the small target detection layer which is developed by increasing the number of detection layers from three to four. The experimental results on multiple object detection dataset show that the proposed model achieves significant improvement in mean average precision (mAP) of almost 98% for various types of targets and an average precision (AP) of nearly 99% for small targets on the other hand it achieves 5.19% (mAP) and 9.79% (AP) compared to the YOLOv5s model. Furthermore, comparative analysis with other models like YOLOv7 and YOLOv8 shows that the proposed model has superior performance in terms of object detection.

VegaEdge: Edge AI confluence for real-time IoT-applications in highway safety

Traditional highway safety and monitoring solutions, reliant on surveillance cameras, face limitations due to their dependence on high-speed internet connectivity and the remote processing of Artificial Intelligence (AI) algorithms. This reliance introduces latency, undermining the real-time detection and analysis crucial for highway applications. The fusion of AI with the Internet of Things (IoT) opens new avenues for highway safety and surveillance innovation. Yet, most existing solutions are confined to vehicle detection and tracking, hindered by edge-IoT platforms’ limited power and processing capabilities. Addressing these limitations, this paper presents VegaEdge, an AI framework optimized for edge-IoT devices capable of real-time vehicle detection and tracking, trajectory forecasting, and identifying anomalous driving behaviors, such as road departures, sudden stops, and hazardous merges. A novel lightweight anomaly detection algorithm based on trajectory prediction is used for identifying hazardous driving on highways. VegaEdge demonstrates its versatility and efficiency across various traffic conditions and roadway configurations and has been evaluated on platforms like the Nvidia Jetson Orin and Xavier NX. The Nvidia Jetson Orin processes up to 738 trajectories per second and detects up to 140 vehicles in a single frame. Additionally, the Carolinas Anomaly Dataset (CAD) an extension of the Carolinas Highway Dataset (CHD) is introduced. While CHD consists of standard highway vehicle videos and trajectories, CAD includes video data of anomalous driving behaviors, providing a crucial resource for enhancing anomaly detection algorithms. CAD is available at https://github.com/TeCSAR-UNCC/Carolinas_Dataset#chd-anomaly-test-set.

Determining the number of small-sized radars in a network with coherent signal processing for the detection of stealth aerial vehicles

The object of this study is the process of determining the number of small-sized radars in the network when detecting stealth unmanned aerial vehicles. The main hypothesis of the study assumed that determining the optimal number of small-sized radars in the network will make it possible not to waste unnecessary resources of radars to detect stealth unmanned aerial vehicles. The main stages of detection of a stealth unmanned aerial vehicle by a network of small-sized radars are: – reception of the signal reflected from a stealth unmanned aerial vehicle by all small-sized radars of the network; – coordinated filtering of incoming signals in each small-sized radar; – compensation of phase shifts in each matched filter; – coherent addition of output signals from each matched filter at the output of the receivers of each of the N small-sized radars performing reception; – formation of a complex bypass at the output of the corresponding Doppler channel in each small-sized radar of the network; – coherent processing of signals from all elements of the network of small-sized radars; – detection of the output signal from the adder of coherent signals. At the same time, compensation for the random initial phase of signals reflected from a stealth unmanned aerial vehicle is also performed. It has been established that the increase in the elements of the network of small-sized radars increases the value of the conditional probability of correct detection. Such an increase is more significant when the number of elements in the network of small-sized radars is increased to two or three. The gain in the signal/noise ratio when adding elements to the network of small-sized radars was evaluated. It was established that the optimal number of small-sized radars in a network with coherent signal processing when detecting stealth unmanned aerial vehicles is 2‒3 radars

Vehicle Classification and Counting for Traffic Analysis based on Single-stage YOLOv8 Model

Intelligent traffic systems are emerging and becoming part of smart city infrastructure that requires computer vision applications to provide traffic information like vehicle classification and counting in real-time. Vehicle counting helps detect heavy traffic on roads, and vehicle classification helps enforce further processes like speed estimation to enforce speed limit laws based on the vehicle class. Deep learning computer vision-based systems provide automatic feature extractions that are robust to changes in lighting, shadows, and occlusions. This paper proposes a software solution for a real-time traffic monitoring system based on a cutting-edge single-stage deep learning model through the state-of-the-art YOLOv8 algorithm. YOLOv8 is the most recent model of the YOLO family, which provides object detection and classification through its CNN architecture. The proposed work detects vehicles and counts them based on their class. The four common vehicle classes are sedan cars, buses, trucks, and motorcycles, and a counter for each class is displayed on the system’s output screen in real-time and recorded in a log file. The results of the proposed system running on the Nvidia GTX 1070 GPU show an average accuracy of 96.58% with an average error of 3.42% for vehicle detection and an average accuracy of 97.54% with a 2.46% average error for vehicle counting. For vehicle classification, the results for the four vehicle classes (car, bus, truck, and motorcycle) show an accuracy of (94.7%, 94.7%, 96.2%, 99.7%), precision (95%, 100%, 81.4%, 100%), recall (97.9%, 36.3%, 100%, 66.6%), and the f1-score (96.3%, 53.2%, 89.7%, 79.9%), respectively. Index Terms— Computer Vision, Deep Learning, Vehicle Detection, Traffic Analysis, YOLO.

Snow-CLOCs: Camera-LiDAR Object Candidate Fusion for 3D Object Detection in Snowy Conditions.

Although existing 3D object-detection methods have achieved promising results on conventional datasets, it is still challenging to detect objects in data collected under adverse weather conditions. Data distortion from LiDAR and cameras in such conditions leads to poor performance of traditional single-sensor detection methods. Multi-modal data-fusion methods struggle with data distortion and low alignment accuracy, making accurate target detection difficult. To address this, we propose a multi-modal object-detection algorithm, Snow-CLOCs, specifically for snowy conditions. In image detection, we improved the YOLOv5 algorithm by integrating the InceptionNeXt network to enhance feature extraction and using the Wise-IoU algorithm to reduce dependency on high-quality data. For LiDAR point-cloud detection, we built upon the SECOND algorithm and employed the DROR filter to remove noise, enhancing detection accuracy. We combined the detection results from the camera and LiDAR into a unified detection set, represented using a sparse tensor, and extracted features through a 2D convolutional neural network to achieve object detection and localization. Snow-CLOCs achieved a detection accuracy of 86.61% for vehicle detection in snowy conditions.

Documentation and management of complex 3D morphologies through digital technology

Digital technologies offer an effective solution for documenting and managing complex three-dimensional morphologies, particularly in the context of Underground Built Heritage (UBH). Our research explores novel digitization strategies by employing various technological tools for 3D scanning. We focus on a selected case study: the Grotta di San Michele Arcangelo. Range-based techniques using LiDAR-UAV (Light Detection and Ranging set in Unmanned Aerial Vehicles) were used for the cave’s exterior, integrated with SLAM (Simultaneous Localization and Mapping) technology was employed for the cave’s interior and close-range photogrammetry techniques captured intricate wall decorations. The results yielded precise geometric information crucial for subsequent three-dimensional modelling of architectural and natural spaces. Furthermore, these techniques were integrated to enhance information management, facilitate conservation efforts, and update existing documentation. The resulting comprehensive database is an essential tool for understanding, monitoring, and effectively managing these intricate underground environments.

Integration of YOLOv5 Algorithm and OpenCV in Innovative Smart Parking Management Approach

The problem of automatic parking lot identification and vehicle detection in open areas is becoming increasingly important due to the increase in the number of vehicles in Indonesia, particularly in big cities, resulting in difficulties in finding parking spaces during peak hours. In this condition, drivers often have to compete for parking spaces. This research aims to develop a smart parking system that integrates YOLOv5 and OpenCV algorithms. This approach thoroughly combines both algorithms to identify parking spaces and detect vehicles in real time in various parking scenarios. It is carried out in an open area with reference to parking conditions at the BRIN Bandung office. This study collected data from three different parking lot conditions, namely empty, partially occupied, and full. In each condition, the system successfully detected the parking lots and vehicles accurately. The novel contribution of this research is the development of a smart parking system that uses an integrated approach, providing an effective solution to the challenges of parking lot availability and vehicle detection. Using the advantages of both algorithms, we successfully created a system that can identify parking spaces and detect vehicles accurately and efficiently under various parking circumstances. Therefore, this research makes a significant contribution to the development of smart and adaptive parking management technology.

Visual-based vehicle detection with adaptive oversampling

In this study, the class imbalance issue in vehicle detection was addressed. Specifically, certain classes such as Tow Truck were found to have significantly fewer samples compared to others such as normal trucks. This imbalance could be adversely impacted algorithm performance, favouring abundant classes over underrepresented ones. After thorough analysis, an adaptive dataset augmentation approach was proposed for the underrepresented classes. Evaluation was first performed on classic and state-of-the-art object detection methods. All experiments were undertaken on a tiny dataset called Multimedia University Diversity Dataset (MMUVD). The fastest training process and the highest mean average precision (mAP), which stood at 0.686 for mAP50 and 0.439 for mAP50-95, were demonstrated by You Only Look Once version 8 nano (YOLOv8n). By applying adaptive oversampling to the dataset and retesting it again on YOLOv8n, mAP50 was improved to 0.950 and mAP50-95 to 0.717, respectively. Notably, the contribution lay in identifying the optimal detection algorithm for vehicle detection, and the proposed adaptive oversampling method ensured consistent performance across all classes, enhancing the overall accuracy and reliability of the system.

Self-supervised network for low-light traffic image enhancement based on deep noise and artifacts removal

In the intelligent transportation system (ITS), detecting vehicles and pedestrians in low-light conditions is challenging due to the low contrast between objects and the background. Recently, many works have enhanced low-light images using deep learning-based methods, but these methods require paired images during training, which are impractical to obtain in real-world traffic scenarios. Therefore, we propose a self-supervised network (SSN) for low-light traffic image enhancement that can be trained without paired images. To avoid amplifying noise and artifacts in the processed image during enhancement, we first proposed a denoising net to reduce the noise and artifacts in the input image. Then the processed image can be enhanced by the enhancement net. Considering the compression of the traffic image, we designed an artifacts removal net to improve the quality of the enhanced image. We proposed several effective and differential losses to make SSN trainable with low-light images only. To better integrate the extracted features from different levels in the network, we also proposed an attention module named the multi-head non-local block. In experiments, we evaluated SSN and other low-light image enhancement methods on two low-light traffic image sets: the Berkeley Deep Drive (BDD) dataset and the Hong Kong night-time multi-class vehicle (HK) dataset. The results indicated that SSN significantly improves upon other methods in visual comparison and some blind image quality metrics. We also conducted comparisons on classical ITS tasks like vehicle detection on the images enhanced by SSN and other methods, which further verified its effectiveness.

Faster R-CNN based on frame difference and spatiotemporal context for vehicle detection

Faster R-CNN based on frame difference and spatiotemporal context for vehicle detection

An intelligent FL-based vehicle route optimization protocol for green and sustainable IoT connected IoV

The intelligent Internet of Vehicles (IoV) provides superior results in effectively addressing complex transportation challenges. Predicting vehicle traffic, crashes, demand, location, communication, and travel safety are all critical issues in today’s transportation systems. The proposed paper optimizes vehicle traffic by incorporating reroute recommendations, increasing the use of public transportation, and providing onboard vehicle drivers with intelligent health assistance using Federated Learning (FL). This research also focuses on resolving complex transportation issues such as a vehicle’s current location, exact vehicle count information on each route, and onboard vehicle vacant seat information. Furthermore, vehicle communication contributes to the proposed system’s efficiency by avoiding communication delays or information loss to registered users and the cloud server. An intelligent FL-based scheme for vehicle route optimization has been proposed as part of this research to prevent vehicle traffic in a real-time Internet of Things (IoT) connected IoV transportation system. The vehicle detection approach determines the number of vehicles traveling on each route to recommend the best route to registered users. The effective implementation of cluster-based vehicle communication and location estimation models enhances the efficiency of the proposed system.

Automatic target detection and recognition of military vehicles in synthetic aperture radar images is fostered by optimizing VGG-googLeNet with the giraffe kicking optimization algorithm

Automatic target detection and recognition of military vehicles in synthetic aperture radar images is fostered by optimizing VGG-googLeNet with the giraffe kicking optimization algorithm

Fractional B-Spline Wavelets and U-Net Architecture for Robust and Reliable Vehicle Detection in Snowy Conditions.

This paper addresses the critical need for advanced real-time vehicle detection methodologies in Vehicle Intelligence Systems (VIS), especially in the context of using Unmanned Aerial Vehicles (UAVs) for data acquisition in severe weather conditions, such as heavy snowfall typical of the Nordic region. Traditional vehicle detection techniques, which often rely on custom-engineered features and deterministic algorithms, fall short in adapting to diverse environmental challenges, leading to a demand for more precise and sophisticated methods. The limitations of current architectures, particularly when deployed in real-time on edge devices with restricted computational capabilities, are highlighted as significant hurdles in the development of efficient vehicle detection systems. To bridge this gap, our research focuses on the formulation of an innovative approach that combines the fractional B-spline wavelet transform with a tailored U-Net architecture, operational on a Raspberry Pi 4. This method aims to enhance vehicle detection and localization by leveraging the unique attributes of the NVD dataset, which comprises drone-captured imagery under the harsh winter conditions of northern Sweden. The dataset, featuring 8450 annotated frames with 26,313 vehicles, serves as the foundation for evaluating the proposed technique. The comparative analysis of the proposed method against state-of-the-art detectors, such as YOLO and Faster RCNN, in both accuracy and efficiency on constrained devices, emphasizes the capability of our method to balance the trade-off between speed and accuracy, thereby broadening its utility across various domains.