Online Tracking Research Articles

A Vision Detection Scheme Based on Deep Learning in a Waste Plastics Sorting System

The preliminary sorting of plastic products is a necessary step to improve the utilization of waste resources. To improve the quality and efficiency of sorting, a plastic detection scheme based on deep learning is proposed in this paper for a waste plastics sorting system based on vision detection. In this scheme, the YOLOX (You Only Look Once) object detection model and the DeepSORT (Deep Simple Online and Realtime Tracking) multiple object tracking algorithm are improved and combined to make them more suitable for plastic sorting. For plastic detection, multiple data augmentations are combined to improve the detection effect, while BN (Batch Normalization) layer fusion and mixed precision inference are adopted to accelerate the model. For plastic tracking, the improved YOLOX is used as a detector, and the tracking effect is further improved by optimizing the deep cosine metric learning and the metric in the matching stage. Based on this, virtual detection lines are set up to filter and extract information to determine the sorted objects. The experimental results show that the scheme proposed in this paper makes full use of vision information to achieve dynamic and real-time detection of plastics. The system is effective and versatile for sorting complex objects.

Usability and Enforceability of Global Privacy Control

Web tracking by ad networks and other data-driven businesses is often privacy-invasive. Privacy laws, such as the California Consumer Privacy Act, aim to give people more control over their data. In particular, they provide a right to opt out from web tracking via privacy preference signals, notably Global Privacy Control (GPC). GPC holds the promise of enabling people to exercise their opt out rights on the web. Broad adoption of GPC hinges on its usability. In a usability survey we find that 94% of the participants would turn on GPC indicating a need for such efficient and effective opt out mechanism. 81% of the participants in our survey also have a correct understanding of what GPC does ensuring that their intent is accurately represented by their choice. The effectiveness of GPC is dependent on whether websites' GPC compliance can be enforced. A site's GPC compliance can be analyzed based on privacy flags, such as the US Privacy String, which is used on many sites to indicate the opt out status of a web user. Leveraging the US Privacy String for GPC purposes we implement a proof-of-concept browser extension that successfully and correctly analyzes sites' GPC compliance at a rate of 89%. We further implement a web crawler for our browser extension demonstrating that our analysis approach is scalable. We find that many sites do not respect GPC opt out signals despite being legally obligated to do so. Only 54/464 (12%) sites with a US Privacy String opt out users after having received a GPC signal.

Real-time tracking and dynamic berthing information extraction system with 2D LiDAR data

In smart ports, accurately perceiving the state of a berthing vessel relative to the berth is essential for ensuring the safety and efficiency of vessels and ports. This paper presents a real-time tracking and dynamic berthing information extraction system based on two-dimensional light detection and ranging (LiDAR). First, data preprocessing, fast segmentation, and filtering are performed in polar coordinates based on the continuous measurements and high angular resolution of the LiDAR. Moreover, the ship clusters are merged owing to “blind spot”. Then, simple online and real-time tracking with ship features is proposed by modifying the simple online and real-time tracking with a deep association metric framework to remove invalid frames and adapt to different berthing scenarios. Qualitative and quantitative analyses of the dynamic distance and approach angle berthing parameters relative to the dock are performed. The frequency of the calculation results can reach a rate of 20 Hz for LiDAR. On-site monitoring experiments were conducted on the training ship YuKun berthing/unberthing in the Dalian Port and a simulated berthing scenario was also established to verify the algorithm. The proposed method proved feasible and effective for dynamic ship target tracking and ensuring safe ship berthing.

Deep learning pose estimation for multi-cattle lameness detection

The objective of this study was to develop a fully automated multiple-cow real-time lameness detection system using a deep learning approach for cattle detection and pose estimation that could be deployed across dairy farms. Utilising computer vision and deep learning, the system can analyse simultaneously both the posture and gait of each cow within a camera field of view to a very high degree of accuracy (94–100%). Twenty-five video sequences containing 250 cows in varying degrees of lameness were recorded and independently scored by three accredited Agriculture and Horticulture Development Board (AHDB) mobility scorers using the AHDB dairy mobility scoring system to provide ground truth lameness data. These observers showed significant inter-observer reliability. Video sequences were broken down into their constituent frames and with a further 500 images downloaded from google, annotated with 15 anatomical points for each animal. A modified Mask-RCNN estimated the pose of each cow to output 5 key-points to determine back arching and 2 key-points to determine head position. Using the SORT (simple, online, and real-time tracking) algorithm, cows were tracked as they move through frames of the video sequence (i.e., in moving animals). All the features were combined using the CatBoost gradient boosting algorithm with accuracy being determined using threefold cross-validation including recursive feature elimination. Precision was assessed using Cohen’s kappa coefficient and assessments of precision and recall. This methodology was applied to cows with varying degrees of lameness (according to accredited scoring, n = 3) and demonstrated that some characteristics directly associated with lameness could be monitored simultaneously. By combining the algorithm results over time, more robust evaluation of individual cow lameness was obtained. The model showed high performance for predicting and matching the ground truth lameness data with the outputs of the algorithm. Overall, threefold lameness detection accuracy of 100% and a lameness severity classification accuracy of 94% respectively was achieved with a high degree of precision (Cohen’s kappa = 0.8782, precision = 0.8650 and recall = 0.9209).

Improving maritime traffic surveillance in inland waterways using the robust fusion of AIS and visual data

To guarantee vessel traffic safety in inland waterways, the automatic identification system (AIS) and shore-based cameras have been widely adopted to monitor moving vessels. The AIS data could provide the unique maritime mobile service identity (MMSI), position coordinates (i.e., latitude and longitude), course over ground, and speed over ground for the vessels of interest. In contrast, the cameras could directly display the visual appearance of vessels but fail to accurately grasp the vessels’ identity information and motion parameters. In this paper, we propose to improve the maritime traffic surveillance in inland waterways using the robust fusion of AIS and visual data. It is able to obtain more accurate vessel tracking results and kinematic characteristics. In particular, to robustly track the visual vessels under complex scenarios, we first propose an anti-occlusion vessel tracking method based on the simple online and real-time tracking with a deep association metric (DeepSORT) method. We then preprocess and predict the vessel positions to obtain synchronous AIS and visual data. Before the implementation of AIS and visual data fusion, the AIS position coordinates in the geodetic coordinate system will be projected into the image coordinate system via the coordinate transformation. A multi-feature similarity measurement-based Hungarian algorithm is finally proposed to robustly and accurately fuse the AIS and visual data in the image coordinate system. For the sake of repeating fusion experiments, we have also presented a new multi-sensor dataset containing AIS data and shore-based camera imagery. The quantitative and qualitative experiments show that our fusion method is capable of improving the maritime traffic surveillance in inland waterways. It can overcome the vessel occlusion problem and fully utilizes the advantages of multi-source data to promote the maritime surveillance, resulting in enhanced vessel traffic safety and efficiency. In this work, the presented multi-sensor dataset and source code are available at https://github.com/QuJX/AIS-Visual-Fusion.

Data‐based learning control for optimization of nonlinear systems

Data‐based learning control for optimization of nonlinear systems

Consumer Privacy Choices and (Un)Targeted Advertising Along the Purchase Journey

Advertising to a consumer provides potentially useful information to the consumer and moves them along the purchase journey, and tracking the consumer's online activities enables an advertiser to infer the consumer's purchase journey state and target repeat ads accordingly. However, many consumers dislike being tracked, and, furthermore, repeat advertising may lead to ad wearout. The authors develop a model with consumers, an advertiser, and an ad network to investigate, under the preceding considerations, the impact of regulations that endow consumers with the choice to opt in to or out of online tracking. The authors find that, if ad effectiveness is intermediate, opting in to tracking decreases ad repetition; otherwise, opting in increases ad repetition. To make an opt-in decision, a consumer weighs the cost of ad wearout from repeat ads against the benefit of obtaining potentially relevant product information from them, and the consumer opts in to tracking if either ad effectiveness is intermediate or sensitivity to ad wearout is low. This opt-in pattern creates counterintuitive implications; for instance, higher ad effectiveness, even though it implies higher ad valuation for the advertiser, may reduce repeat ads and the ad network's profit. Under regulation that requires consumer consent for tracking, the results shed light on when and why consumers give such consent, and provide useful insights for practitioners and policy makers.

Multi-Optimization Objective Online Tracking-Based Parameter Self-Tuning Method for Sensorless PMSM Drives

It is important to optimize control parameters for the high-performance operation of position sensorless permanent magnet synchronous motor (PMSM) drives. In this article, a multi-optimization objective online tracking (MOOT)-based parameter self-tuning method is proposed to improve the sensorless control performance when the high-frequency signal injection (HFSI) method is used, which involves both offline and online stages. By the parameter self-learning and the model analysis, the injection signal, the position observer, the speed, and current regulators are preliminarily designed offline to ensure that the system can start smoothly. Based on the parameter characteristics analysis, an adaptive law with the speed ripple and torque ripple as the optimization index is designed to realize the online tracking of control and observation parameters. To solve the issues of tuning delay and mistuning, a moving window-based error detection strategy is investigated, which improves the parameter self-tuning speed and accuracy. The proposed method can improve the steady- and transient-state performance and realize the sensorless debug-free operation for HFSI-based method. The effectiveness is verified by experiments on a 2.2-kW PMSM drive platform.

Switch and Refine: A Long-Term Tracking and Segmentation Framework

In long-term video object tracking (VOT) tasks, most long-term trackers are modified from short-term trackers, which contain more and more machine learning modules to improve their performance. However, we empirically find that more modules do not necessarily lead to better results. In this paper, we make the long-term tracking framework simple by carefully selecting the cutting-edge trackers. Specifically, we propose a new long-term VOT framework that combines the benefits of two mainstream short-term tracking pipelines, i.e., the discriminative online tracker and the one-shot Siamese tracker, with a global re-detector awakened when the target is lost. Such a framework fully exploits existing advanced works from three complementary perspectives. Experimental results show that by exploiting the capabilities of existing methods instead of designing new neural networks, we can still achieve remarkable results on seven long-term VOT datasets. By introducing a continuous adjustable speed control parameter, our tracker reaches 20+FPS with only a small performance loss. The refine module not only improves the bounding box estimations but also outputs segmentation masks, so that our framework can handle the video object segmentation (VOS) tasks by using only VOT trackers. We obtain a trade-off between time and accuracy on two representative VOS datasets by only using bounding boxes as the initial input.

Learning to Track Multiple Radar Targets with Long Short-Term Memory Networks

Radar multitarget tracking in a dense clutter environment remains a complex problem to be solved. Most existing solutions still rely on complex motion models and prior distribution knowledge. In this paper, a new online tracking method based on a long short-term memory (LSTM) network is proposed. It combines state prediction, measurement association, and trajectory management functions in an end-to-end manner. We employ LSTM networks to model target motion and trajectory associations, relying on their strong learning ability to learn target motion properties and long-term dependence of trajectory associations from noisy data. Moreover, to address the problem of missing appearance information of radar targets, we propose an architecture based on the LSTM network to calculate similarity function by extracting long-term motion features. And the similarity is applied to trajectory associations to improve their robustness. Our proposed method is validated in simulation scenarios and achieves good results.

Do (Not!) Track Me: Relationship Between Willingness to Participate and Sample Composition in Online Information Behavior Tracking Research

This paper offers a critical look at the promises and drawbacks of a popular, novel data collection technique—online tracking—from the point of view of sample composition. Using data from two large-scale studies about political attitudes and information consumption behavior carried out in Germany and Switzerland, we find that the likelihood of participation in a tracking study at several critical dropout points is systematically related to the gender, age, and education of participants, with men, young, and more educated participants being less likely to dropout of the studies. Our findings also show that these patterns are incremental, as changes in sample composition accumulate over successive study stages. Political interest and ideology were also significantly related to the likelihood of participation in tracking research. The study explores some of the most common concerns associated with tracking research leading to non-participation, finding that they also differ across demographic groups. The implications of these findings are discussed.

Online tracking and prediction of slip ring degradation using chaos theory based on LSTM neural network

The online monitoring of the slip ring is important for ensuring normal operations of wind turbine equipment. A current-carrying friction experiment was conducted to simulate the degradation process of the slip ring. The chaotic parameter enclosing the radius and statistical parameter root mean square (RMS) were used to characterize the multi-sensor signals comprehensively. A new health indicator (HI) was proposed to evaluate the degradation state of slip rings based on long- and short-term memory neural networks. It was fused by the signals of friction vibration, friction torque, voltage and electric current. The HI presents a better prediction effect by the prediction model. At the severe stage of the slip ring, the evaluation criteria mean absolute error, root mean square error and mean percentage error of the HI were 0.0306, 0.0323 and 5.0225% respectively. These values are much better than the RMS of the vibration signal. The results verify that the method can effectively determine the real-time degradation state of the slip ring.

A Study of an Online Tracking System for Spark Images of Abrasive Belt-Polishing Workpieces.

During the manual grinding of blades, the workers can estimate the material removal rate based on their experiences from observing the characteristics of the grinding sparks, leading to low grinding accuracy and low efficiency and affecting the processing quality of the blades. As an alternative to the recognition of spark images by the human eye, we used the deep learning algorithm YOLO5 to perform target detection on spark images and obtain spark image regions. First the spark images generated during one turbine blade-grinding process were collected, and some of the images were selected as training samples, with the remaining images used as test samples, which were labelled with LabelImg. Afterwards, the selected images were trained with YOLO5 to obtain an optimisation model. In the end, the trained optimisation model was used to predict the images of the test set. The proposed method was able to detect spark image regions quickly and accurately, with an average accuracy of 0.995. YOLO4 was also used to train and predict spark images, and the two methods were compared. Our findings show that YOLO5 is faster and more accurate than the YOLO4 target detection algorithm and can replace manual observation, laying a specific foundation for the automatic segmentation of spark images and the study of the relationship between the material removal rate and spark images at a later stage, which has some practical value.

An Implementation Methodology for the Online Tracking Simulation Architecture of the 660 MW USCFB

This paper studies the implementation strategy of an online tracking simulation system for the 660 MW ultra supercritical circulating fluidized bed boiler (USCFB). The establishment of the online system can be divided into offline and online phases. In the offline phase, the high-precision model and initial condition database of the boiler are first established. After the initial conditions are loaded, the initial conditions are adjusted through mass and energy compensation to make the initial state of the model consistent with the current operating state of the unit. In the online phase, this paper analyzes the global sensitivity of the 660 MW USCFB, establishes an online tracking mechanism based on the PI controller, and conducts simulation tests. The experiment shows that the tracking simulation system established in this paper can track the analog signal quickly and accurately.

Video-based marker-free tracking and multi-scale analysis of mouse locomotor activity and behavioral aspects in an open field arena: A perspective approach to the quantification of complex gait disturbances associated with Alzheimer's disease.

Complex gait disturbances represent one of the prominent manifestations of various neurophysiological conditions, including widespread neurodegenerative disorders such as Alzheimer's and Parkinson's diseases. Therefore, instrumental measurement techniques and automatic computerized analysis appears essential for the differential diagnostics, as well as for the assessment of treatment effectiveness from experimental animal models to clinical settings. Here we present a marker-free instrumental approach to the analysis of gait disturbances in animal models. Our approach is based on the analysis of video recordings obtained with a camera placed underneath an open field arena with transparent floor using the DeeperCut algorithm capable of online tracking of individual animal body parts, such as the snout, the paws and the tail. The extracted trajectories of animal body parts are next analyzed using an original computerized methodology that relies upon a generalized scalable model based on fractional Brownian motion with parameters identified by detrended partial cross-correlation analysis. We have shown that in a mouse model representative movement patterns are characterized by two asymptotic regimes characterized by integrated 1/f noise at small scales and nearly random displacements at large scales separated by a single crossover. More detailed analysis of gait disturbances revealed that the detrended cross-correlations between the movements of the snout, paws and tail relative to the animal body midpoint exhibit statistically significant discrepancies in the Alzheimer's disease mouse model compared to the control group at scales around the location of the crossover. We expect that the proposed approach, due to its universality, robustness and clear physical interpretation, is a promising direction for the design of applied analysis tools for the diagnostics of various gait disturbances and behavioral aspects in animal models. We further believe that the suggested mathematical models could be relevant as a complementary tool in clinical diagnostics of various neurophysiological conditions associated with movement disorders.

Eco-Driving of General Mixed Platoons With CAVs and HDVs

Eco-driving has been widely investigated over the last decade, but most studies focused on an individual vehicle or a vehicle platoon consisting of pure connected and automated vehicles (CAVs). Recently, mixed vehicle platoons consisting of both CAVs and human-driven vehicles (HDVs) have attracted much interest, considering the fact that HDVs will mix with CAVs in the traffic system for a long period. This paper proposes an eco-driving strategy for mixed platoons, composing of both offline planning and online tracking. In offline planning, an energy-efficient speed reference and a gearshift reference are determined by using the characteristics of each vehicle and future traffic information through dynamic programming. Offline planning optimised the vehicle speed and gearshift to allow the vehicle powertrain working at a high efficiency region. In online tracking, two different types of model predictive controls (MPCs) are proposed to control the CAVs in real-time. The MPCs are designed to achieve precise speed reference tracking performance and guarantee platoon string stability, respectively. Meanwhile, HDVs within the mixed platoon can be located anywhere in the platoon except working as the first vehicle to improve flexibility. Therefore, the proposed eco-driving strategy is applicable to mixed platoons with more general structures in ordering. The key contribution of this study is that the proposed eco-driving strategy can optimise the total fuel consumption for general mixed platoons. Simulation results show that the proposed eco-driving strategy improves the fuel economy of a mixed platoon by up to 6.39% compared to the benchmark conventional-adaptive cruise control strategies.

Track geometry prediction for Laser Metal Deposition based on on-line artificial vision and deep neural networks

Laser Metal Deposition (LMD) is an additive manufacturing technology that attracts great interest from the industry, thanks to its potential to realize parts with complex geometries in one piece, and to repair damaged ones, while maintaining good mechanical properties. Nevertheless, the complexity of this process has limited its widespread adoption, since different part geometries, strategies and boundary conditions can yield very different results in terms of external shapes and inner flaws. Moreover, monitoring part quality during the process execution is very challenging, as direct measurements of both structural and geometrical properties are mostly impracticable. This work proposes an on-line monitoring and prediction approach for LMD that exploits coaxial melt pool images, together with process input data, to estimate the size of a track deposited by LMD. In particular, a novel deep learning architecture combines the output of a convolutional neural network (that takes melt pool images as inputs) with scalar variables (process and trajectory data). Various network architectures are evaluated, suggesting to use at least three convolutional layers. Furthermore, results imply a certain degree of invariance to the number and size of dense layers. The effectiveness of the proposed method is demonstrated basing on experiments performed on single tracks deposited by LMD using powders of Inconel 718, a relevant material for the aerospace and automotive sectors. • LMD process requires on-line control for achieving predictable results. • Matrix and scalar variables are mixed in the proposed deep network structure. • Geometry prediction by AI is tested on single-track experiments. • Image information extracted by the CNN improves on-line track size prediction.

Fast Timeline Based Multi Object Online Tracking

Abstract Fast state-of-the-art multi-object-tracking (MOT) schemes, such as reported in challenges MOT16 and Mot20, perform tracking on a single sensor, often couple tracking and detection, support only one kind of object representation or don’t take varying latencies and update rates into account. We propose a fast generic MOT system for use in real world applications which is capable of tracking objects from different sensor / detector types with their respective latencies and update rates. An SORT inspired online tracking scheme is extended by time awareness using timelines as unifying concept. The system supports different object, sensor and filter and tracking types by modularizing and generalizing the online tracking scheme, while ensuring high performance using an efficient data-oriented C++-template-based implementation. Using the proposed system we achieve, with comparable evaluation metrics, framerates up to ten times higher than the fastest MOT schemes publicly listed for the axis-aligned bounding-box tracking challenges MOT17 and MOT20.

Data-driven design of the Belle II track segment finder

The Belle II experiment relies on a level-1 trigger system to reduce noise background and preselect events of interest for particle physics. The Central Drift Chamber is the main track detector which makes its trigger system important for online track reconstruction. To improve its hit efficiency, an extension of the track segment finder for low angle tracks is proposed. By combining hardware and software development flows, an automated data-driven pipeline is created and three different-sized hardware concepts are implemented. The operation point is adjustable to balance hit efficiency against hit purity in the trigger system.

Pedestrian walking speed monitoring at street scale by an in-flight drone.

The walking speed of pedestrians is not only a reflection of one's physiological condition and health status but also a key parameter in the evaluation of the service level of urban facilities and traffic engineering applications, which is important for urban design and planning. Currently, the three main ways to obtain walking speed are based on trails, wearable devices, and images. The first two cannot be popularized in larger open areas, while the image-based approach requires multiple cameras to cooperate in order to extract the walking speed of an entire street, which is costly. In this study, a method for extracting the pedestrian walking speed at a street scale from in-flight drone video is proposed. Pedestrians are detected and tracked by You Only Look Once version 5 (YOLOv5) and Simple Online and Realtime Tracking with a Deep Association Metric (DeepSORT) algorithms in the video taken from a flying unmanned aerial vehicle (UAV). The distance that pedestrians traveled related to the ground per fixed time interval is calculated using a combined algorithm of Scale-Invariant Feature Transform (SIFT) and random sample consensus (RANSAC) followed by a geometric correction algorithm. Compared to ground truth values, it shows that 90.5% of the corrected walking speed predictions have an absolute error of less than 0.1 m/s. Overall, the method we have proposed is accurate and feasible. A particular advantage of this method is the ability to accurately predict the walking speed of pedestrians without keeping the flight speed of the UAV constant, facilitating accurate measurements by non-specialist technicians. In addition, because of the unrestricted flight range of the UAV, the method can be applied to the entire scale of the street, which assists in a better understanding of how the settings and layouts of urban affect people's behavior.