Car Detection Research Articles

Vehicle Crash Detection using YOLO Algorithm

Car accidents cause a large number of deaths and disabilities every day, a certain proportion of which result from untimely treatment and secondary accidents. To some extent, automatic car accident detection can shorten response time of rescue agencies and vehicles around accidents to improve rescue efficiency and traffic safety level. —Road Accidents, a very common reason of tragic deaths and many times the victim dies due to non-reporting of such accidents to the proper authority. Since the accident was not reported the lack of emergency medical care results in death. We live in an era of technology where we are moving towards making the city, A Smart City. A smart city with smart AI based traffic monitoring and reporting mechanism can help providing medical emergencies in real time and this would result in saving lots of life. Traditional Traffic systems are equipped with IP cameras and sensors, and are already installed in most part of the city to monitor and control traffic. These systems are able to generate traffic tickets automatically. In this paper we are proposing a more advanced traffic monitoring system which can identify and detect moving objects such are cars, bikes etc in live camera feeds and detect collision of these moving objects and immediately send emergency alerts to the nearby authority for them to take necessary actions.

A Machine Learning Framework for Automated Accident Detection Based on Multimodal Sensors in Cars.

Identifying accident patterns is one of the most vital research foci of driving analysis. Environmental or safety applications and the growing area of fleet management all benefit from accident detection contributions by minimizing the risk vehicles and drivers are subject to, improving their service and reducing overhead costs. Some solutions have been proposed in the past literature for automated accident detection that are mainly based on traffic data or external sensors. However, traffic data can be difficult to access, while external sensors can end up being difficult to set up and unreliable, depending on how they are used. Additionally, the scarcity of accident detection data has limited the type of approaches used in the past, leaving in particular, machine learning (ML) relatively unexplored. Thus, in this paper, we propose a ML framework for automated car accident detection based on mutimodal in-car sensors. Our work is a unique and innovative study on detecting real-world driving accidents by applying state-of-the-art feature extraction methods using basic sensors in cars. In total, five different feature extraction approaches, including techniques based on feature engineering and feature learning with deep learning are evaluated on the strategic highway research program (SHRP2) naturalistic driving study (NDS) crash data set. The main observations of this study are as follows: (1) CNN features with a SVM classifier obtain very promising results, outperforming all other tested approaches. (2) Feature engineering and feature learning approaches were finding different best performing features. Therefore, our fusion experiment indicates that these two feature sets can be efficiently combined. (3) Unsupervised feature extraction remarkably achieves a notable performance score.

Research on Crack Detection Algorithm of Mining Car Baffle

Introduction: Image processing technology is widely used for crack detection. This technology is used to build a data acquisition system and uses computer vision technology for image analysis. Because of its simplicity in processing, many of the image processing detection methods were proposed. It is relatively easy to deploy and has a low cost. Method: The heterogeneity of the external light usually changes the authenticity of each target in the image, which will seriously cause the experiment to fail. At this time, the image needs to be processed by the gamma transform. Based on the analysis of the characteristics of the image of the mine car baffle, this paper improves the Gamma transform and uses the it to enhance the image. Result: It can be concluded that the algorithm in this paper can accurately detect crack areas with an actual width greater than 1.2 mm, and the error between the detected crack length and the actual length is between (-2, 2) mm. In practice, this error is completely acceptable. Discussion: To compare the performance of a new crack detection method with the existing methods. The two most well-known traditional methods, Canny and Sobel edge detection, are selected. Although the Sobel edge detection provides some crack information. The texture of the surface of the mine cart baffle detected has caused great interference to the crack identification. Conclusion: If the cracks appearing on the mine car baffle are not found in time, they often cause accidents. Therefore, effective crack detection must be performed. If a manual inspection is adopted for crack detection, it will be labor-intensive and easy to miss inspection. In order to reduce the labor of crack detection of mine cars and improve the accuracy of detection, this paper, based on the detection platform built, performs pre-processing, image enhancement, and convolution operations on the collected crack images of the mine car baffle.

Car detection and damage segmentation in the real scene using a deep learning approach

Car detection and damage segmentation in the real scene using a deep learning approach

Detection of Logos of Moving Vehicles under Complex Lighting Conditions

This study proposes a method for vehicle logo detection and recognition to detect missing and inaccurate vehicle marks under complex lighting conditions. For images acquired in complex light conditions, adaptive image enhancement is used to improve the accuracy of car sign detection by more than 2%; for the problems of multi-scale and detection speed of vehicle logo recognition in different images, the paper improves the target detection algorithm to improve the detection accuracy by more than 3%. The adaptive image enhancement algorithm and improved You Only Look One-level Feature (YOLOF) detection algorithm proposed in this study can effectively improve the correct identification rate under complex lighting conditions.

Development of metro track geometry fault diagnosis convolutional neural network model based on car-body vibration data

Traditional track geometry fault diagnosis methods, such as track geometry car, track detection trolley and manual measurement, suffer from low detection frequency. Therefore, using vibration data from in-service metro vehicles to diagnose metro track geometry faults can improve the efficiency of metro track condition detection. Hence, a metro track geometry fault diagnosis model using car-body vibration was established in this study based on a convolutional neural network model, and the effects of the structural hyperparameters on the model performance were analysed. First, the problem of car-body vibration data collection was addressed, and a mileage-locating model without the global navigation satellite system was developed. Then, the metro track geometry fault diagnosis model was derived using a convolutional neural network whose architecture is determined by four hyperparameters. Finally, a case study was performed using the Beijing Subway 1 track geometry car detection data and car-body vibration data. The results demonstrated the effectiveness of the proposed model.

A LiDAR–Camera Fusion 3D Object Detection Algorithm

3D object detection with LiDAR and camera fusion has always been a challenge for autonomous driving. This work proposes a deep neural network (namely FuDNN) for LiDAR–camera fusion 3D object detection. Firstly, a 2D backbone is designed to extract features from camera images. Secondly, an attention-based fusion sub-network is designed to fuse the features extracted by the 2D backbone and the features extracted from 3D LiDAR point clouds by PointNet++. Besides, the FuDNN, which uses the RPN and the refinement work of PointRCNN to obtain 3D box predictions, was tested on the public KITTI dataset. Experiments on the KITTI validation set show that the proposed FuDNN achieves AP values of 92.48, 82.90, and 80.51 at easy, moderate, and hard difficulty levels for car detection. The proposed FuDNN improves the performance of LiDAR–camera fusion 3D object detection in the car category of the public KITTI dataset.

Research on Both the Classification and Quality Control Methods of the Car Seat Backrest Based on Machine Vision

In order to solve the problems of slow manual inspection speed and low fault detection accuracy of car seat back parts, this article using Q company’s car seat back parts researches and designs a car seat back classification and quality inspection screening system. Firstly, SURF (speeded up robust features) is combined with the CNN (convolutional neural network) to classify three types of car seat backrests: A, B, and C. Then, to establish the spring hook angle detection model of the car seat back to detect the misfitting and omission of the Class A car seat back springs, experimental results showed that the neural network-based car seat back detection method proposed in this paper had a feature point mismatch rate, which is less than 1.5% in the classification and recognition of car seat backs. The recognition rate of the training sample was 100% and that of the test sample was 99.56%. The accuracy rate of detection when inspecting 50 car seat backrests reached 98%, and the test results showed that the system can effectively reduce labor costs and improve the detection efficiency of auto parts.

A Multiple‐State Ion Synaptic Transistor Applicable to Abnormal Car Detection with Transfer Learning

An artificial synapse is an essential element to construct a hardware‐based artificial neural network (ANN). While various synaptic devices have been proposed along with studies on electrical characteristics and proper applications, a small number of conductance states with nonlinear and asymmetric conductance changes have been problematic and imposed limits on computational performance. Their applications are thus still limited to the classification of simple images or acoustic datasets. Herein, a polymer electrolyte‐gated synaptic transistor (pEGST) is demonstrated for video‐based learning and inference using transfer learning. In particular, abnormal car detection (ACD) is attempted with video‐based learning and inference to avoid traffic accidents. The pEGST showed multiple states of 8,192 (=13 bits) for weight modulation with linear and symmetric conductance changes and helped reduce the error rate to 3% to judge whether a car in a video is abnormal.

Lightweight convolutional neural network for real-time 3D object detection in road and railway environments

For smart mobility, and autonomous vehicles (AV), it is necessary to have a very precise perception of the environment to guarantee reliable decision-making, and to be able to extend the results obtained for the road sector to other areas such as rail. To this end, we introduce a new single-stage monocular real-time 3D object detection convolutional neural network (CNN) based on YOLOv5, dedicated to smart mobility applications for both road and rail environments. To perform the 3D parameter regression, we replace YOLOv5’s anchor boxes with our hybrid anchor boxes. Our method is available in different model sizes such as YOLOv5: small, medium, and large. The new model that we propose is optimized for real-time embedded constraints (lightweight, speed, and accuracy) that takes advantage of the improvement brought by split attention (SA) convolutions called small split attention model (Small-SA). To validate our CNN model, we also introduce a new virtual dataset for both road and rail environments by leveraging the video game Grand Theft Auto V (GTAV). We provide extensive results of our different models on both KITTI and our own GTAV datasets. Through our results, we show that our method is the fastest available 3D object detection with accuracy results close to state-of-the-art methods on the KITTI road dataset. We further demonstrate that the pre-training process on our GTAV virtual dataset improves the accuracy on real datasets such as KITTI, thus allowing our method to obtain an even greater accuracy than state-of-the-art approaches with 16.16% 3D average precision on hard car detection with inference time of 11.1 ms/image on an RTX 3080 GPU.

Flexible neural network for fast and accurate road scene perception

Accurate object detection on the road is the most important requirement of autonomous vehicles. Extensive work has been accomplished for car, pedestrian, and cyclist detection; however, comparatively, very few efforts have been put into 2D object detection. In this article, a dynamic approach is investigated to design a perfect unified neural network that could achieve the best results based on our available hardware. The proposed architecture is based on CSPNet for feature extraction in an end-to-end way. The net extracts visual features by using backbone subnet, visual object detection is based on a feature pyramid network (FPN). In order to increase the net flexibility, an auto-anchor generating method is applied to the detection layer that makes the net suitable for any datasets. For fine-tuning the net, activation, optimization, and loss functions are considered along with multiple check points. The proposed net is trained and tested based on the benchmark KITTI dataset. Our extensive experiments show that the proposed model for visual object detection is superior to others, where other nets output very low accuracy for pedestrian and cyclist detection, our proposed model achieves 99.3% recall rate based on our dataset.

Moving Vehicle Detection and Recognition Technology based on Artificial Intelligence

The existing target detection and recognition technology has the problem of fuzzy features of moving vehicles, which leads to poor detection effect. A moving car detection and recognition technology based on artificial intelligence is designed. The point operation is adopted to enhance the high frequency information of the image, increase the image contrast, and delineate the video image tracking target. The motion vector similarity is used to predict the moving target area in the next frame of the image. The texture features of the moving car are extracted by artificial intelligence, and the center moment is calculated by the gray histogram distribution curve, the edge feature extraction algorithm is used to set the detection and recognition mode. Experimental results: under complex conditions, this design technology, compared with the other two kinds of moving vehicle detection and recognition technology, detected three more moving vehicles, which proved that the application prospect of the moving vehicle detection and recognition technology integrated with artificial intelligence is broader.

Car, Cyclist and Pedestrian Object Detection Based on YOLOv5

The target detection of car, cyclist and pedestrian is of great significance to the realization of automatic monitoring and artificial intelligence assisted driving system. In order to rapidly detect car, cyclist and pedestrian, You Only Look Once (YOLO) model is applied to car, cyclist and pedestrian detection tasks in this paper. Experimental result shows that among the four versions of YOLOv5, the mean average precision (MAP) is more than 91.7%, and YOLOv5x has the best recognition ability, which reaches 92.6%. Among the three categories, car recognition accuracy is the highest, followed by cyclist and pedestrian.

Moving target detection in elevator hoistway with millimeter-wave radar

Detection of the elevator car and counterweight is most important for the safety operation in the elevator maintenance. The purpose of this research is the development of the moving target alert system with millimeter-wave radar, which can detect the velocity and position of the moving target such as the elevator car and counterweight. In this report, we benchmarked the sensing method for detecting the moving target in the elevator hoistway and designed the moving target alert system. We developed the prototype measurement system with the 77-GHz millimeter-wave radar. Measured results of detecting the car and counterweight showed that the developed millimeter-wave radar set in the elevator pit can detect the car and the counterweight in the maintenance speed (15 m/min) and normal speed (105 m/min) within 10 m of the elevator pit.

Automatic Feature Engineering-Based Optimization Method for Car Loan Fraud Detection

Fraud detection is one of the core issues of loan risk control, which aims to detect fraudulent loan applications and safeguard the property of both individuals and organizations. Because of its close relevance to the security of financial operations, fraud detection has received widespread attention from industry. In recent years, with the rapid development of artificial intelligence technology, an automatic feature engineering method that can help to generate features has been applied to fraud detection with good results. However, in car loan fraud detection, the existing methods do not satisfy the requirements because of overreliance on behavioral features. To tackle this issue, this paper proposed an optimized deep feature synthesis (DFS) method in the automatic feature engineering scheme to improve the car loan fraud detection. Problems like feature dimension explosion, low interpretability, long training time, and low detection accuracy are solved by compressing abstract and uninterpretable features to limit the depth of DFS algorithm. Experiments are developed based on actual car loan credit database to evaluate the performance of the proposed scheme. Compared with traditional automatic feature engineering methods, the number of features and training time are reduced by 92.5% and 54.3%, respectively, whereas accuracy is improved by 23%. The experiment demonstrates that our scheme effectively improved the existing automatic feature engineering car loan fraud detection methods.

ENSEMBLE NEURO-FUZZY BASED SYSTEM FOR VEHICLE THEFT PREDICTION AND RECOVERY

Vehicle theft is continuously being reported as a global prevalent crime. It often aids the perpetuation of other related crimes such as kidnapping, armed robbery, terrorism and human trafficking. The traditional mode of combating vehicle theft crime is faced with abnormallies hindering accurate, timely prediction and recovery of stolen vehicles from criminals. This paper presents a computational Artificial Intelligence (A.I) technique known as Ensemble Neuro-Fuzzy modeled system with the aim of minimizing investigation time and number of deployed security operatives towards achieving a high successful rate in the prediction, detection and recovery of stolen cars. A collection of data collected from the Criminal Investigation Department of the Nigeria Police Force, were further analyzed through Dimensionality Reduction formula and Routine Activity Approach (RTA) to extract the most significant features. Dataset were sub-divided into 60, 20 and 20% for training, testing and validating the model respectively. A significant result of 92.91% obtained with this model showed that it is most efficient in predicting, detecting and recovering of stolen vehicles as compared with other machine learning algorithms such as Random Tree, Naïve Bayes, J48 and Decision Rule of prediction accuracy of 86.51%, 71.24%, 67.68% and 55.73% respectively.

A semi-supervised 3D object detection method for autonomous driving

3D object detection is very important for autonomous driving, but most of algorithms heavily rely on data annotation, which is a laborious job. Semi-supervised learning is an effective strategy to reduce the heavy annotation burden in this task. In this paper, we propose a semi-supervised 3D object detection method. To reduce the workload of 3D annotations, we adopt the teacher-student framework to generate pseudo-labels from unlabeled training data, and use a label filtering method to improve the pseudo label quality. For the convenience of deployment, we introduce the simple and fast one-stage object detector. We validate our method on the KITTI dataset. For moderate Car detection task, our method can achieve 76.28 mAP using half labels compared with 77.34 mAP of the PointPillars using all labels. Experiment results show that our method can also achieve approximate performance on Pedestrian and Cyclist detection task with fewer labels.

Sensitivity and Specificity of CD19.CAR-T Cell Detection by Flow Cytometry and PCR.

Chimeric-antigen-receptor-T (CAR-T) cells are currently revolutionizing the field of cancer immunotherapy. Therefore, there is an urgent need for CAR-T cell monitoring by clinicians to assess cell expansion and persistence in patients. CAR-T cell manufacturers and researchers need to evaluate transduction efficiency and vector copy number for quality control. Here, CAR expression was analyzed in peripheral blood samples from patients and healthy donors by flow cytometry with four commercially available detection reagents and on the gene level by quantitative polymerase chain reaction (qPCR). Flow cytometric analysis of CAR expression showed higher mean CAR expression values for CD19 CAR detection reagent and the F(ab’)2 antibody than Protein L and CD19 Protein. In addition, the CD19 CAR detection reagent showed a significantly lower median background staining of 0.02% (range 0.007–0.06%) when compared to the F(ab’)2 antibody, CD19 protein and Protein L with 0.80% (range 0.47–1.58%), 0.65% (range 0.25–1.35%) and 0.73% (range 0.44–1.23%). Furthermore, flow cytometry-based CAR-T cell frequencies by CD19 CAR detection reagent showed a good correlation with qPCR results. In conclusion, quality control of CAR-T cell products can be performed by FACS and qPCR. For the monitoring of CAR-T cell frequencies by FACS in patients, CAR detection reagents with a low background staining are preferable.

Kte-X19 in Relapsed or Refractory Mantle-Cell Lymphoma, a “Real-Life” Study from the Descar-T Registry and Lysa Group

Kte-X19 in Relapsed or Refractory Mantle-Cell Lymphoma, a “Real-Life” Study from the Descar-T Registry and Lysa Group

CD19.CAR-T Cell Analysis Using Flow Cytometry and PCR

Introduction: After approval of CD19.CAR-T cell therapy by both FDA and EMA, chimeric antigen receptor T (CAR-T) cell therapy has been established as a new and innovative therapy method for patients with relapsed/refractory acute lymphoblastic leukemia (ALL) and non-Hodgkin lymphoma (NHL). Because of this, there is great need for reliable methods for CAR-T cell monitoring by clinicians to further assess cell expansion, distribution and persistence in patients. This is both necessary for CAR-T cell manufacturing sites and researchers to assess transduction efficiency and vector copy numbers per cell. Therefore, we analyzed CAR expression of CD19.CAR-T cells using flow cytometry as well as quantitative polymerase chain reaction (qPCR).Methods: In this study, four different commercially available CD19.CAR-T cell staining reagents (Protein L, CD19 protein, F(ab')2 antibody, and CD19.CAR-T cell detection reagent) were used to evaluate CAR-T cell products and peripheral blood samples of both patients and healthy donors by flow cytometry. Furthermore, qPCR was performed with data comparison using flow cytometry results.Results: Flow cytometric analysis of CAR expression showed higher mean CAR expression values for CD19 CAR detection reagent and the F(ab')2 antibody than Protein L and CD19 Protein. Moreove, the CD19 CAR detection reagent showed a significantly lower median background staining of 0.02% (range 0.007-0.06%) when compared to F(ab')2 antibody, CD19 protein and Protein L with 0.80% (range 0.47-1.58%.), 0.68% (range 0.30-1.38%) and 0.73% (range 0.44-1.23%). Furthermore, flow cytometry-based CAR-T cell frequencies by CD19 CAR detection reagent showed a good correlation with qPCR results.Conclusion: CAR-T staining was successfully performed with all tested commercially available CAR detection reagents. Evaluation of manufactured CAR-T cells as well as quality control was comparably done using FACS and qPCR. Because of lower frequencies of CAR-T cells in the patient samples, CAR-T cell staining reagents with a low background staining should be preferably used. DisclosuresSauer: Abbvie: Consultancy, Speakers Bureau; Pfizer: Consultancy, Speakers Bureau; Matterhorn Biosciences AG: Consultancy, Other: DSMB/SAB Member; Takeda: Consultancy, Other: DSMB/SAB Member. Schubert: Gilead: Consultancy. Müller-Tidow: Janssen: Consultancy, Research Funding; Bioline: Research Funding; Pfizer: Research Funding. Schmitt: TolerogenixX: Current holder of individual stocks in a privately-held company; Novartis: Other: Travel grants, Research Funding; MSD: Membership on an entity's Board of Directors or advisory committees; Kite Gilead: Other: Travel grants; Apogenix: Research Funding; Hexal: Other: Travel grants, Research Funding; Bluebird Bio: Other: Travel grants. Schmitt: Hexal: Other: Travel grant; Therakos/Mallinckrodt: Research Funding; Jazz Pharmaceuticals: Other: Travel grant; TolerogenixX Ltd: Current Employment.