Surveillance Tasks Research Articles

ALFormer: Attribute localization transformer in pedestrian attribute recognition

Pedestrian attribute recognition is an important task for intelligent video surveillance. However, existing methods struggle to accurately localize discriminative regions for each attribute. We propose Attribute Localization Transformer (ALFormer), a novel framework to improve spatial localization through two key components. First, we introduce Mask Contrast Learning (MCL) to suppress regional feature relevance, forcing the model to focus on intrinsic spatial areas for each attribute. Second, we design an Attribute Spatial Memory (ASM) module to generate reliable attention maps that capture inherent locations for each attribute. Extensive experiments on two benchmark datasets demonstrate state-of-the-art performance of ALFormer. Ablation studies and visualizations verify the effectiveness of the proposed modules in improving attribute localization. Our work provides a simple yet effective approach to exploit spatial consistency for enhanced pedestrian attribute recognition.

The Influence Of Understanding The Collision Regulations (COLREG) 1972 On Competence Of Navigation Supervision Tasks On Ships

This research examined the impact of understanding the 1972 Collision Regulations (COLREG) on navigational watch competence on ships. COLREG, established by the International Maritime Organization (IMO), provides essential guidelines to prevent sea accidents, which every crew member must know. Mastery of COLREG is crucial for cadets in Nautical programs, ensuring they perform effectively during Marine Practice and as future officers. This study utilized a quantitative approach, with structured research objectives, subjects, objects, samples, and data sources. Methodology spanned from data collection to analysis. Findings indicated significant direct correlations: understanding of COLREG with navigation watch duty competence (coefficient = 0.201, p < 0.05), application of COLREG rules (coefficient = 0.235, p < 0.05), situation analysis (coefficient = 0.154, p < 0.05), and taking appropriate actions (coefficient = 0.188, p < 0.05). Similarly, cadet performance showed significant correlations with understanding and applying COLREG rules, situation analysis, and taking appropriate actions, with coefficients ranging from 0.078 to 0.174 (all p < 0.05). Additionally, a strong link was found between cadet performance and navigation watch duty competence (coefficient = 0.458, p < 0.05). The results of the study showed that the competence of navigation surveillance tasks on board ships was closely related to the factors of understanding the 1972 colreg regulations. The findings demonstrate that seafarers with a strong grasp of COLREG can more effectively identify navigational priorities and make appropriate decisions and actions in emergency situations.

Computer Vision-Based Intelligent Traffic Surveillance: Multi-Vehicle Tracking and Detection

The application of vehicle detection in real-time traffic surveillance systems is one of the challenging research fields with different objectives. One of the problems is the detection of many vehicles simultaneously in a video sequence sourced from CCTV cameras. In many works, the focus is only on detecting vehicle classes such as motorcycles, buses, trucks, and cars or special vehicles such as ambulances and others. In this research, we propose to apply 13 classes of vehicle types and implement YOLOv4 in the traffic surveillance task. More specifically, all classes are labeled, and then the YOLOv4 model is trained on 800 images and tested on 23 videos from three intersections in Medan City, namely Juanda Katamso Intersection, Gatot Subroto Intersection, and Uniland Intersection. Based on the test results, YOLOv4 proves successful in detecting many vehicles in frame-by-frame sequence with various types of vehicles. All vehicle detection data will be stored in the file.

A Review on the Identification and Classification of Activities for the Purpose of Tracking Objects in Video Surveillance

An increasing number of academics are becoming interested in the field of image processing since object tracking in video surveillance is a significant application and a growing subject of research in the field of technique known as video tracking is one that involves the utilization of a camera in order to find a moving object or many moving things throughout the course of time. As a result of its key qualities, video surveillance serves a variety of functions, including but not limited to human-computer interactions, security and surveillance, video communication, traffic management, and public areas such as airports, underground stations, and large-scale events, amongst others. Keeping track of a target in a congested area is still a difficult video surveillance task. The processing framework for video surveillanceis completed by a sequential flow that includes the detection of moving objects, their classification, tracking, and behavior identification. This study examines tracking techniques, classifies them into many categories, and concentrates on significant and practical tracking techniques. This paper discusses several tracking systems, including active contour-based, area-based, and others, along with their advantages and disadvantages. Various tracking techniques are discussed along with thorough descriptions. Finally, researchers presentan overview of potential research areas after reviewing general tactics and a scan of the literature on various methodologies.

Пути совершенствования эпидемиологического надзора за эндемичными биогельминтозами Дальнего Востока

One of the tasks of epidemiological surveillance of parasitic diseases is to maintain sanitary and epidemiological well-being for endemic biohelminthiasis in the territories included in the range of pathogens. The inclusion of a person in the circulation of pathogens depends on a complex of social factors, primarily on the nutritional characteristics of residents. Long-term studies show that the official registration data do not fully reflect the real picture of the population affected by pathogens of biohelminthiasis. Sanitary and parasitological examination of fish occupies one of the leading places in the system of preventive measures. The main task at the present stage is to provide a differentiated approach to the development and optimization of means and methods of disease prevention among the population of endemic territories, depending on the type of focus, the level of endemia and, accordingly, the degree of risk of infection of the population. Key words: clonorhiasis, nanofietosis, metagonimosis, epidemiological surveillance.

Addressing Information Inequality for Text-Based Person Search via Pedestrian-Centric Visual Denoising and Bias-Aware Alignments

Text-based person search is an important task in video surveillance, which aims to retrieve the corresponding pedestrian images with a given description. In this fine-grained retrieval task, accurate cross-modal information matching is an essential yet challenging problem. However, existing methods usually ignore the information inequality between modalities, which could introduce great difficulties to cross-modal matching. Specifically, in this task, the images inevitably contain some pedestrian-irrelevant noise like background and occlusion, and the descriptions could be biased to partial pedestrian content in images. With that in mind, in this paper, we propose a Text-Guided Denoising and Alignment (TGDA) model to alleviate the information inequality and realize effective cross-modal matching. In TGDA, we first design a prototype-based denoising module, which integrates pedestrian knowledge from textual features into a prototype vector and uses it as guidance to filter out pedestrian-irrelevant noise from visual features. Thereafter, a bias-aware alignment module is introduced, which guides our model to focus on the description-biased pedestrian content in cross-modal features consistently. Through extensive experiments, the effectiveness of both modules has been validated. Besides, our TGDA achieves state-of-the-art performance on various related benchmarks.

Optimal tasks assignment policy in multi-task multi-attempt missions

Motivated by real-world applications like unmanned reconnaissance aerial vehicles, this paper considers a multi-task multi-attempt mission system, where each task may be attempted multiple times and each attempt may be performed simultaneously by multiple components to enhance the task completion probability. Such an active redundancy, on the other hand, incurs high cost and high risk associated with the failures of the components. To balance the reward and the risk, this paper formulates and solves a new optimization problem, which determines the number of components performing each uncompleted task in each attempt, referred to as the task assignment policy (TAP), minimizing the expected mission cost (EMC). A recursive algorithm is proposed to evaluate the EMC (aggregating the expected operational cost, cost of lost components, and cost associated with uncompleted tasks) for the considered multi-task multi-attempt system under a given TAP. Based on the suggested EMC evaluation algorithm, the genetic algorithm is implemented to solve the optimal TAP problem. A detailed case study of an unmanned reconnaissance aerial vehicle system performing five independent surveillance tasks is conducted to examine the impacts of several model parameters on the EMC, task successful completion probabilities, and the optimal TAP solutions.

UAV remote sensing image stitching via improved VGG16 Siamese feature extraction network

UAV remote sensing image stitching can provide a more comprehensive and continuous area by stitching several overlapping region images. It has been widely used in many fields, such as airborne reconnaissance and surveillance tasks, agriculture production, geological disaster monitoring, and so on. However, traditional feature-based image stitching methods often require hand-crafted features and multiple execution steps, which could extract many redundant feature points in non-overlapping regions and may fail to stitch images with weak textures. To solve this problem, we propose an improved VGG16 Siamese network for the UAV remote sensing image stitching model to achieve end-to-end stitching. The first 13 layers of the VGG16 network were intercepted to form a shared weights feature extraction network, and an improved Squeeze-and-Excitation module was introduced to effectively extract the features of the overlapping areas of the images. A network of regressing affine matrix was designed by using the LeakyReLu activation function to protect the relevant feature maps after feature fine matching, which improves the accuracy of image stitching. Besides, we constructed datasets to train our network: a single UAV remote sensing image was used to form a training image pair based on a limited degree of affine transformation, and the real label of affine transformation between the image pairs was obtained. We compared the SIFT+RANSAC, APAP, and deep homography estimation-based image stitching algorithm on the UAV remote sensing datasets. Experiments show the value of structural similarity(SSIM) increased by 24.1%, and the root mean square error(RMSE) was reduced by 14.69%. Moreover, the proposed stitching model also has good stitching results for UAV remote sensing images as well as weak texture images in subjective visual effects.

Ship detection based on YOLO algorithm for visible images

AbstractShip detection is a crucial task for waterway surveillance and channel optimization, especially in close proximity to the shore. However, detecting ship in visible image‐based detection remains a challenge due to the limited nature of visible image datasets. To address this issue, the Inland Ships Data Set (ISDS) is constructed to facilitate research on ship identification. On the other hand, most detection methods struggle to accurately identify ships that are small in size. Therefore, a visible image‐based ship detection model is proposed that employs a multi‐scale weighted feature fusion structure with the YOLOv4 detection model to improve the efficacy of small ship detection. Specifically, the YOLOv4 model is improved through fusing multi‐scale feature, redesigning priori frame, and enhancing loss function. The model, named YOLOv4‐MSW (i.e. YOLOv4 based on Multi‐Scale Weighted feature fusion), exhibits improved performance on ship detection in experiments conducted on the ISDS dataset, outperforming the original YOLOv4 model by improving the average precision (AP) by 4.87% and the recall rate by 10.03%. Meanwhile, the model achieve better detection accuracy and improve the average precision rate by at least 0.86% compared to existing learned object detection methods. The code related to this work are released athttps://github.com/Sunhuashan/YOLOv4‐MSW. The whole dataset is available athttps://drive.google.com/drive/folders/1fzJ2fcqiko6lFwqIEGghMceoQgv‐8jBy.

Poster] ADS-B anomaly detection in the surveillance of low-altitude aircrafts

In the past few years, the fast increase in air traffic load has brought new challenges for air traffic controllers. The air surveillance task has become harder and as a consequence, the actual monitoring tools need to be improved. In this work, a method based on deep learning that automatically detects ADS-B spoofing attacks is proposed. As autonomous drone technologies will, in the near future, be more and more developed, this study focuses on low-altitude traffic. Our tool is based on a classifier model that raises anomalies between true aircraft trajectory shapes and supposed aircraft categories (e.g. planes, helicopters). The proposed approach can detect spoofing attacks with a success rate of 96.2%.

Low-Cost Computer-Vision-Based Embedded Systems for UAVs

Unmanned Aerial Vehicles (UAVs) are versatile, adapting hardware and software for research. They are vital for remote monitoring, especially in challenging settings such as volcano observation with limited access. In response, economical computer vision systems provide a remedy by processing data, boosting UAV autonomy, and assisting in maneuvering. Through the application of these technologies, researchers can effectively monitor remote areas, thus improving surveillance capabilities. Moreover, flight controllers employ onboard tools to gather data, further enhancing UAV navigation during surveillance tasks. For energy efficiency and comprehensive coverage, this paper introduces a budget-friendly prototype aiding UAV navigation, minimizing effects on endurance. The prototype prioritizes improved maneuvering via the integrated landing and obstacle avoidance system (LOAS). Employing open-source software and MAVLink communication, these systems underwent testing on a Pixhawk-equipped quadcopter. Programmed on a Raspberry Pi onboard computer, the prototype includes a distance sensor and basic camera to meet low computational and weight demands.Tests occurred in controlled environments, with systems performing well in 90% of cases. The Pixhawk and Raspberry Pi documented quad actions during evasive and landing maneuvers. Results prove the prototype’s efficacy in refining UAV navigation. Integrating this cost-effective, energy-efficient model holds promise for long-term mission enhancement—cutting costs, expanding terrain coverage, and boosting surveillance capabilities.

A Review of the Various Control Algorithms for Trajectory Control of Unmanned Underwater Vehicles

Unmanned underwater vehicles (UUVs) have become increasingly popular in recent years due to their use in various applications. The motivations for using UUVs include the exploration of difficult and dangerous underwater environments, military tasks in mine detection, intelligence gathering and surveillance, the inspection of offshore oil and gas infrastructure in the oil and gas industry, scientific research for studying marine life, and the search and rescue of missing persons or submerged airplanes or boats in underwater environments. UUVs offer many advantages in achieving the desired applications with increased safety, efficiency, and cost-effectiveness. However, there are also several challenges associated with their communication, navigation, power requirements, maintenance, and payload limitations. These types of vehicles are also prone to various disturbances caused by currents of the ocean, propulsion systems, and unmolded uncertainties. Practically, it is a challenging task to design a controller that will ensure optimal performance under these conditions. Therefore, the control system design is of prime importance in the overall development of UUVs. Also, the UUV controller receives input from different sensors, and the data from these sensors are used by the controller to perform different tasks. The control systems of UUVs should take into account all uncertainties and make them stable so that all sensors can perform optimally. This paper presents a complete review of different control system design algorithms for UUVs. The basic logic designs of several control system algorithms are also presented. A comparison is made based on reliability, robustness, precession, and the ability of the controller to handle the nonlinearity that is faced by UUVs during their missions. Simulation and experimental results are thoroughly studied to gain insight into each algorithm. The advantages and disadvantages of each algorithm are also presented, which will facilitate the selection of a suitable algorithm for the control system design of UUVs.

Prefab-Net: Segmentation of prefabricated building from satellite imagery with deep neural network

The development of deep learning techniques such as convolutional neural networks (CNN) has injected new vitality into the study of semantic segmentation. Benefiting from the development of new technology, the automatic and intelligent work of remote sensing image segmentation has also entered a new stage. With the further improvement of China’s urbanization rate, urban construction is also developing rapidly. One of the most important tasks in land surveillance is to monitor buildings above ground using remote sensing images. Prefabricated building is the most common type of illegal building in developing area. At present, supervision mainly relies on the traditional human-centered method, which is time-consuming and low in efficiency. Whether the deep learning methods can benefit the prefabricated building segmentation has not yet been tested. By taking the realistic demand into account, this paper proposes a customized neural network structure, which combines multiple refined modules and improves the feature extraction, especially for the prefabricated building segmentation task. Through the comparison with the baseline algorithm, the specifically refined model for the prefabricated building has shown better results in our experimental area, which proves the feasibility and superiority in practical application.

Impacts of AI-Generated Confidence and Explanations on Task Performance and Trust in Human-autonomy Teaming

In human-autonomy teaming (HAT), human operators and intelligent agents cooperate and coordinate together to achieve shared goals. This study aimed to enhance autonomy transparency by proposing the automatic generation of confidence and explanations (bounding boxes, and bounding boxes and keypoints) and investigating their impacts in HAT. A total of 36 participants engaged in a simulated surveillance task, during which they were assisted by the intelligent agents we designed using Keypoint Faster R-CNN. As a result, we found that visual explanations using bounding boxes and keypoints improved detection task performance only when confidence was not visualized. Moreover, participants had higher trust in and preference for autonomy when visual explanations were provided but whether confidence was visualized did not influence their trust and preference. These findings have implications for the design of autonomy and can facilitate human-machine interactions in HAT.

Dynamic Weighted Gradient Reversal Network for Visible-infrared Person Re-identification

Due to intra-modality variations and cross-modality discrepancy, visible-infrared person re-identification (VI Re-ID) is an important and challenging task in intelligent video surveillance. The cross-modality discrepancy is mainly caused by the differences between visible images and infrared images, the inherent essence of which is heterogeneous. To alleviate this discrepancy, we propose a Dynamic Weighted Gradient Reversal Network (DGRNet) to enhance the learning of discriminative common representations by confusing the modality discrimination. In the proposed DGRNet, we design the gradient reversal model guiding adversarial training between identity classifier and modality discriminator to reduce the modality discrepancy of the same person in different modalities. Furthermore, we propose an optimization training method, that is, designing dynamic weight of gradient reversal to achieve optimal adversarial training, and dynamic weight has the ability to dynamically and adaptively evaluate the significance of target loss term, without involving hyper-parameter tuning. Extensive experiments were conducted on two public VI Re-ID datasets, SYSU-MM01 and RegDB. The experimental results show that the proposed DGRNet outperforms state-of-the-art methods and demonstrate the effectiveness of the DGRNet to learn more discriminative common representations for VI Re-ID.

Explainable Algorithm to Predict Passenger Flow at Cincinnati/Northern Kentucky International Airport

We present an algorithm for the prediction of the incoming passengers at the airport’s security checkpoint with a resolution of 15 min for 2 weeks following the day of the prediction. This is characterized not only for its performance but also from its explainability in the outcomes. The algorithm has been integrated successfully at Cincinnati/Northern Kentucky International Airport (CVG) for daily passenger predictions, which ultimately equipped the airport managers with information to perform a data-driven deployment of the necessary airport resources. These range from tailored Transportation Security Administration (TSA) officer schedules and staff allocation to surveillance and supervision tasks. Some of the indirect consequences of this technology are congestion avoidance, the improvement of overall security at the airport facilities, the reduction of the waiting times, and the enhancement of the passenger experience.

Evaluation of Single-Trial Classification to Control a Visual ERP-BCI under a Situation Awareness Scenario.

An event-related potential (ERP)-based brain-computer interface (BCI) can be used to monitor a user's cognitive state during a surveillance task in a situational awareness context. The present study explores the use of an ERP-BCI for detecting new planes in an air traffic controller (ATC). Two experiments were conducted to evaluate the impact of different visual factors on target detection. Experiment 1 validated the type of stimulus used and the effect of not knowing its appearance location in an ERP-BCI scenario. Experiment 2 evaluated the effect of the size of the target stimulus appearance area and the stimulus salience in an ATC scenario. The main results demonstrate that the size of the plane appearance area had a negative impact on the detection performance and on the amplitude of the P300 component. Future studies should address this issue to improve the performance of an ATC in stimulus detection using an ERP-BCI.

Distributed Multi-Target Search and Surveillance Mission Planning for Unmanned Aerial Vehicles in Uncertain Environments

In this paper, a distributed, autonomous, cooperative mission-planning (DACMP) approach was proposed to focus on the problem of the real-time cooperative searching and surveillance of multiple unmanned aerial vehicles (multi-UAVs) with threats in uncertain and highly dynamic environments. To deal with this problem, a time-varying probabilistic grid graph was designed to represent the perception of a target based on its a priori dynamics. A heuristic search strategy based on pyramidal maps was also proposed. Using map information at different scales makes it easier to integrate local and global information, thereby improving the search capability of UAVs, which has not been previously considered. Moreover, we proposed an adaptive distributed task assignment method for cooperative search and surveillance tasks by considering the UAV motion environment as a potential field and modeling the effects of uncertain maps and targets on candidate solutions through potential field values. The results highlight the advantages of search task execution efficiency. In addition, simulations of different scenarios show that the proposed approach can provide a feasible solution for multiple UAVs in different situations and is flexible and stable in time-sensitive environments.

Routing UAVs in landslides Monitoring: A neural network heuristic for team orienteering with mandatory visits

Unmanned aerial vehicles (UAVs) are widely used for surveillance in both civilian and military scenarios. The utilization of UAVs provides an opportunity for monitoring landslide-prone areas by automatically collecting geological information, thereby reducing the risks and the time required to be working in harsh environments. Due to the maximum travel time limit of UAVs and the geographical dispersion of landslide-prone areas, multiple UAVs are dispatched for surveillance tasks, and landslide-prone areas with high emergency priorities require mandatory visits. Here, we investigate a team orienteering problem with mandatory visits (TOPMV) for routing multi-UAVs to monitor scattered landslide-prone areas, with mandatory visits on those in poorly stable states. The proposed TOPMV aims to plan the optimal multi-UAV paths for maximizing the total amount of collected geological information. To solve the TOPMV with a realistic scale, we develop a large neighborhood search (LNS) algorithm embedding a neural network heuristic (NNH), in which the embedded NNH learns to perform adaptive destroy operators through a hierarchical recurrent graph convolutional network (HRGCN). We consider a real-world case study for monitoring of landslide-prone areas in three counties in southern Shaanxi Province, China. Finally, we test the proposed NNH on both small- and large-scale benchmark instances of the team orienteering problem. The experimental results demonstrate that our proposed NNH exhibits higher efficiency and provides better solution quality than state-of-the-art methods, especially in large-scale settings.

Real-Time Workload Estimation Using Eye Tracking: A Bayesian Inference Approach

Workload management is a critical concern in shared control of unmanned ground vehicles. In response to this challenge, prior studies have developed methods to estimate human operators’ workload by analyzing their physiological data. However, these studies have primarily adopted a single-model-single-feature or a single-model-multiple-feature approach. The present study proposes a Bayesian inference model to estimate workload, which leverages different machine learning models for different features. We conducted a human subject experiment with 24 participants, in which a human operator teleoperated a simulated High Mobility Multipurpose Wheeled Vehicle (HMMWV) with the help from an autonomy while performing a surveillance task simultaneously. Participants’ eye-related features, including gaze trajectory and pupil size change, were used as the physiological input to the proposed Bayesian inference model. Results show that the Bayesian inference model achieves a 0.823 F 1 score, 0.824 precision, and 0.821 recall, outperforming the single models.