A Fast and Accurate Real-Time Vehicle Detection Method Using Deep Learning for Unconstrained Environments

In order to improve the speed of road vehicle and lane line detection, a road vehicle and lane line detection method with improved YOLOv4-tiny algorithm is proposed by using K-means++ clustering algorithm instead of the original K-means algorithm. Using the Mosaic data enhancement method, four images are randomly deflated and then stitched into one image to enrich the detection target background. The optimal weight values are derived by multi-scale training using the GPU through image feature extraction of the BDD10K dataset to achieve the detection of vehicle and lane line targets in the images. The results show that the improved YOLOv4-tiny algorithm achieves a detection speed of 134 FPS and an average accuracy of 77.84% mAP in highway lane line detection. After comparison, the detection speed of the improved algorithm is significantly improved, effectively improving the efficiency of highway vehicle and lane line detection.

To realize road traffic flow surveillance under various environments which contain poor visibility conditions, we have already proposed two vehicle detection methods using thermal images taken with an infrared thermal camera. The first method uses pattern recognition for the windshields and their surroundings to detect vehicles. However, the first method decreases the vehicle detection accuracy in winter season. To maintain high vehicle detection accuracy in all seasons, we developed the second method. The second method uses tires' thermal energy reflection areas on a road as the detection targets. The second method did not achieve high detection accuracy for vehicles on left-hand and right-hand lanes except for two center-lanes. Therefore, we have developed a new method based on the second method to increase the vehicle detection accuracy. This paper proposes the new method and shows that the detection accuracy for vehicles on all lanes is 92.1%. Therefore, by combining the first method and the new method, high vehicle detection accuracies are maintained under various environments, and road traffic flow surveillance can be realized.

We present a novel method for vehicle detection and classification in aerial imagery. First, change detection analyzes a pair of mutually aligned images captured at the same location but at different time instants to generate vehicle proposals. Next, a trained Convolutional Neural Network (CNN) classifier is applied to (1) determine if a proposal truly contains a vehicle, and (2) classify any vehicle present into major categories. Experimental results using infrared (IR) data demonstrate the efficacy of our method: vehicle detection rate is over 99%; light-duty vehicles (e.g. sedan) are classified with 89% accuracy, medium-duty vehicles (e.g., van and pickup) with 79% and heavy-duty vehicles (e.g., trucks and buses) with 73% accuracies. The classification performance difference among different vehicle types is attributed to the size of training samples available in each category. Our system can process data at video rate making it feasible for real-time traffic monitoring.

Vehicle detection and tracking plays a major role in military and civilian applications. The accurate detection of multiple vehicles in a complicated traffic environment is too difficult. This process is made more difficult, if there are occlusions between vehicles. The optimal feature selection of a vehicle is also a challenging task. These issues are focused in the existing work namely convolution neural network (CNN) based vehicle detection. However in this work, feature selection is not concentrated and with the lesser feature information classification accuracy is reduced considerably. This is focused and resolved in the proposed research by introducing the method namely Enhanced Convolution neural network with Support Vector Machine (ECNN-SVM) based vehicle detection. In this work initially feature extraction is done. Haar like features are used for this purpose. This has compact representation, capture information from multiple scales, encoded edge and structural information and can be efficient computation. Haar like feature pool is large scale in nature. Enhanced bat optimization is used to select features from this pool. This selection is done by combining sample’s class label and feature value. Confidence level between two thresholds of tracking rectangles is used for classification. Support vector machine (SVM) classifier is combined with local binary pattern to perform this classification. The interference areas between moving objects and vehicles are removed by Enhanced Convolutional Neural Network (ECNN) classifier. The efficiency of the proposed method is computed by using experimental and theoretical analysis.

In the intelligent transportation system, vehicle detection is one of the essential technologies in obstacle avoidance and navigation, however the existing vehicle detection methods cannot meet the actual needs. This paper presents a vehicle detection method combines the intensity and distance information of point cloud, which improves the segmentation performance of nearby objects. Specifically, the data of point cloud collected by lidar is preprocessed first. Then the processed point cloud is clustered by combining its coordinate and intensity information. Finally, the clustered suspected targets are fed to the random forest classifier. Our method can efficiently detect and classify targets in large-scale disordered 3D point cloud with high accuracy. In the real-scanned Livox Mid-40 Lidar dataset, our proposed method improves the detection accuracy by 31% compared with the traditional Euclidean clustering.

Aiming at the problems of low detection and tracking accuracy, poor robustness and inaccurate vehicle flow statistics of multi-vehicle types in surveillance video, an artificial intelligence-based vehicle detection and tracking method was proposed. This method reconstructed the feature extraction network of the target detection model YOLOv5s to enhance the extraction of important features of the target and improve the detection accuracy of the detector. Firstly, Swin Transformer module is introduced into the Backbone network to replace part of C3 modules in the original algorithm, so as to enhance the global modeling ability of the model, better capture the context feature information, and expand the model's receptive field. Then, by comparing different attention mechanisms, NAM attention is selected to access the Neck network to enhance the cross-dimensional interaction between the channel and spatial dimensions of information, reduce information loss and enhance the performance of the network. Finally, the feature extraction network part of the DeepSort algorithm was optimized and retrained on the vehicle rerecognition data set to make it more suitable for vehicle tracking. The experimental results show that the accuracy of vehicle detection by the improved YOLOv5s is increased by 1.93 percentage points compared with the original algorithm. Combined with the DeepSort algorithm, the accuracy of vehicle flow statistics under different lighting conditions such as daytime, evening and night can reach 97.2%, 97.5% and 80.4% respectively.

Intelligent transportation systems (ITS) depend on precise vehicle identification and speed estimation to provide real-time traffic monitoring, accident avoidance, and effective traffic flow management. This paper, a novel deep neural network (DNN)-based method for real-time vehicle detection and speed estimate is presented. For accurate vehicle detection, we use cutting-edge object detection models like YOLOv5, and we use computer vision algorithms to determine a vehicle's speed from video frames. The technology operates reliably in a range of environmental circumstances and processes real-time traffic video streams. High detection accuracy (above 95%) and accurate speed estimation with an average error margin of less than 5 km/h are demonstrated by the experimental findings. A scalable and economical architecture for traffic surveillance applications is provided by the suggested approach.

Vehicle detection in driving simulation using extreme learning machine

Efficient anomaly detection and traffic monitoring are the advance features of Intelligent Transportation Systems (ITS) while existing techniques struggle in handling hazardous scenarios in traffic. Traditional detection methods based on bounding-box approach as discussed in earlier studies, show reduced effectiveness with variable vehicle types and complex traffic patterns deployed in heterogeneous traffic environment. This paper presents an advanced framework of computer vision for real-time traffic monitoring and anomaly detection in different traffic environments. The system utilizes YOLOv8 instance segmentation models that achieves pixel-level accuracy in vehicle detection while maintaining real-time processing capabilities. Using user-defined regions of interest (ROIs) in the paradigm of road-region focused detection eliminates background interference. The proposed framework implements pixel-level vehicle segmentation and supports simultaneous vehicle counting, traffic density analysis and substantially improves accuracy in anomaly detection by eliminating background noise. The proposed system was authenticated on urban traffic datasets across all analyzed frames <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(n=7)$</tex> proving exceptional performance with consistent anomaly detection. The results displayed noteworthy enhancement in processing efficiency and detection accuracy in comparison to existing methodologies, including average occupancy percentages ranging from 63.31 % to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{7 1 \%}$</tex> and maximum occupancy reaching <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{9 3. 7 4 \%}$</tex>.

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.

Transportation plays an important role in urban development However, the vehicle growth in Indonesia is not supported by the number of road. Due to this fact, traffic congestion is easily occurred, especially in big cities. Intelligent Transportation System (ITS) has huge contribution to decrease the traffic congestion. In ITS, vehicle detection is one of challenging issue for traffic surveillance. In this paper, adaptive selective background learning and hole filling algorithm are applied to improve the vehicle detection. The validity of the proposed method is tested by using three scenarios and two parameters. The scenarios are bad weather close range (BW-CR), normal weather close range (NW-CR) and normal weather wide range (NW-WR). While, the parameters are the time duration of stopped vehicle detection and the pixel accuracy. Then, the proposed method (Adaptive Selective Background Learning-Hole Filling algorithm) is compared by another previous vehicle detection method. Generally, the result shows that the proposed method yields a significant improvement in vehicle detection. ASBL-HF can detect the stopped and moved vehicle with free noises. Moreover, ASBL-HF has the best accuracy. The accuracy value is about 98.2%.

When the general preceding vehicle detection is carried for monocular vision, the accuracy of the detection is affected by the dynamic background, illumination and occlusion. This paper proposed a preceding vehicle detection method fusing monocular and binocular vision method. Firstly, the potential area of the preceding vehicle was established by using SSD (Single Shot Multibox Detection) algorithm, then, the binocular vision-based vehicle detection method for the potential area was used, its U-V disparity was calculated, and the accurate preceding vehicle position by the vehicle depth information in the image was obtained; finally, the missed detection and false detections in the monocular vision detector were removed with the U-V disparity detections obtained by binocular vision detector. And the final results combined with the type and position information were outputted. Thereby the accuracy of the preceding vehicle detection was improved. The test result showed that on the condition of KITTI date set the visual fusion method can effectively improve the accuracy of vehicle detection. And compared with SSD algorithm in monocular, the Detection Rate was increased by 1.46%, the Recall Rate was increased by 2.83%, the Missed Rate was reduced by 2.83% and the False Rate was reduced by 1.10% than the monocular SSD algorithm.

In the paper, we adopted an object-oriented image analysis method to detect and classify road vehicles from airborne color digital orthoimagery at a ground pixel resolution of 20cm. Firstly; a vector-generated road mask was used to constrain detection and classification of vehicles to road region. Secondly, image segmentation and edge detection algorithms were performed to separate vehicles from the background in the road region. Then, a fuzzy logic classifier was constructed to classify the extracted object regions into the vehicle and the non-vehicle regions by using the feature information of image objects. Finally, based on the calculated average length and width of vehicles, we classified vehicles into three categories, that is, small, medium and big. And the counts of the three vehicle classes were derived. The automatic counts match manual counts very well.

In order to achieve rapid and accurate detection of vehicle objects in complex traffic conditions, we propose a novel vehicle detection method. Firstly, more contextual and small-object vehicle information can be obtained by our Joint Feature Network (JFN). Secondly, our Evolved Region Proposal Network (EPRN) generates initial anchor boxes by adding an improved version of the region proposal network in this network, and at the same time filters out a large number of false vehicle boxes by soft-Non Maximum Suppression (NMS). Then, our Mask Network (MaskN) generates an example that includes the vehicle occlusion, the generator and discriminator can learn from each other in order to further improve the vehicle object detection capability. Finally, these candidate vehicle detection boxes are optimized to obtain the final vehicle detection boxes by the Fine-Tuning Network(FTN). Through the evaluation experiment on the DETRAC benchmark dataset, we find that in terms of mAP, our method exceeds Faster-RCNN by 11.15%, YOLO by 11.88%, and EB by 1.64%. Besides, our algorithm also has achieved top2 comaring with MS-CNN, YOLO-v3, RefineNet, RetinaNet, Faster-rcnn, DSSD and YOLO-v2 of vehicle category in KITTI dataset.

Wind power prediction using deep neural network based meta regression and transfer learning