People Counting Research Articles

Device-free crowd counting using multi-link WiFi CSI for occupant-driven energy management of HVAC systems

ABSTRACT Accurate estimation of the number of occupants in a building’s thermal zone is a crucial input for the dynamic estimation of the cooling load of a heating, ventilation, and air conditioning equipment. This data, in conjunction with an intelligent control algorithm, can be judiciously employed for enhancing occupant comfort and achieving substantial energy savings through demand-driven operation. In this work, we propose a WiFi channel state information (CSI) based passive people counting model that utilizes a wavelet-based signal denoising scheme and an artificial neural network (ANN) model for automatic feature extraction. The application of the proposed model on a public dataset shows more than 99% accuracy in detecting up to eight people in three rooms of varying sizes. The proposed model was also validated on CSI samples collected from a laboratory environment with five occupants, using ESP32, a low cost and low power IoT device. The results show that the proposed model can perform people counting with an accuracy of up to 98%, using single- and multi-link measurements.

Drone Insights: Unveiling Beach Usage through AI-Powered People Counting

Ocean beaches are a major recreational attraction in many coastal cities, requiring accurate visitor counts for infrastructure planning and value estimation. We developed a novel method to assess beach usage on the Gold Coast, Australia, using 507 drone surveys across 24 beaches. The surveys covered 30 km of coastline, accounting for different seasons, times of day, and environmental conditions. Two AI models were employed: one for counting people on land and in water (91–95% accuracy), and another for identifying usage types (85–92% accuracy). Using drone data, we estimated annual beach usage at 34 million people in 2022/23, with 55% on land and 45% in water—approximately double the most recent estimate from lifeguard counts, which are spatially limited and prone to human error. When applying similar restrictions as lifeguard surveys, drone data estimated 15 million visits, aligning closely with lifeguard counts (within 9%). Temporal (time of day, day of the week, season) and spatial (beach location) factors were the strongest predictors of beach usage, with additional patterns explained by weather variables. Our method, combining drones with AI, enhances the coverage, accuracy, and granularity of beach monitoring, offering a scalable, cost-effective solution for long-term usage assessment.

A new occupancy index model based on artificial vision for enhancing beach management

This study proposes a new method to more effectively plan the use of beaches by combining indices and artificial vision systems. The Overcrowding Index (Iocr) measures the number of people on the beach in relation to its surface area, while the Distancing Index (Idis) evaluates the spatial distribution and distance between beachgoers. Both indices are combined to generate an overall index called the Occupancy Index (Iocu). The proposed methodology uses cameras and computer vision algorithms such as YOLOX and ByteTrack to automate the counting of people and measure distances. This allows for continuous monitoring of the quantity (carrying capacity and density) and distribution of beachgoers (degree of social distancing), as well as a functional prototype in which the indices are calculated in real time. It was observed that as density increased, Iocr showed an inverse trend, being close to 0 when approaching maximum density. The calculation of the distance between groups validated that, even with medium densities, close to the shoreline, the reference distance of 2 m was not accomplished, obtaining a very low Idis (0.18). The resulting Iocu was 0.31, validating the appropriate integration of both indices. Overall, the system's effectiveness for accurately monitoring the number of users and their distribution, and calculating the defined indices for beach management, is demonstrated. The proposed approach provides a valuable tool, allowing a more efficient management of beaches according to their actual occupancy and user distribution.

Video-based automatic people counting for public transport: On-bus versus off-bus deployment

Interest in Automatic People Counting (APC) for crowd detection and management is rapidly growing. While a range of Internet of Things (IoT) sensors and systems exist, video analytics is emerging as a particularly attractive option — especially for applications where more traditional methods of people counting are not available, unreliable or expensive. In this paper we focus on automatic people counting in the public transport context – specifically, rail replacement bus services – in which bus companies are typically contracted to provide bus services to replace train services during periods of planned and unplanned line disruption. This presents a particularly compelling use-case for video-based people counting, while also presenting unique challenges. Field trials are thus vital to the proper assessment of video-based APC solutions, however remain relatively scarce in the literature. While datasets to support research and benchmarking exist, these do not capture the intrinsic complexities of real-world deployment and the implications of selected configurations — in particular, on-vehicle versus off-vehicle use cases. In this paper, we evaluate our own video-based APC solution, representative of state-of-the-art approaches in the literature, in two separate extensive (i.e, multi-day) metropolitan field trials covering both on and off-bus use-cases. Through real-world deployment of the system in both settings, we highlight key differences with respect to APC accuracy, as well as other practical considerations, and the validity of underlying assumptions in both on and off-bus scenarios.

Tasks of visual crowd analysis in intelligent video surveillance systems

Modern urban population growth creates challenges for public safety, in particular due to crowds. This stimulates the development of new crowd management methods that require automated analysis. Visual crowd analysis based on computer vision technologies is a key tool for solving these problems. The development of deep learning has significantly improved the monitoring systems used for urban surveillance, social distancing control, transportation and event management. However, crowd analysis remains challenging due to occlusions, scale variations, unpredictable movement patterns, and complex behavior. To overcome these challenges, new algorithms, models, and large-scale datasets are needed to enable real-time analysis. The main tasks include people counting, object detection, motion analysis, behavior recognition, and anomaly detection. Deep neural networks and transfer learning significantly increase the accuracy and adaptability of such systems, which helps to improve public safety and manage the flow of people.

Crowd Counting Using Meta-Test-Time Adaptation.

Machine learning algorithms are commonly used for quickly and efficiently counting people from a crowd. Test-time adaptation methods for crowd counting adjust model parameters and employ additional data augmentation to better adapt the model to the specific conditions encountered during testing. The majority of current studies concentrate on unsupervised domain adaptation. These approaches commonly perform hundreds of epochs of training iterations, requiring a sizable number of unannotated data of every new target domain apart from annotated data of the source domain. Unlike these methods, we propose a meta-test-time adaptive crowd counting approach called CrowdTTA, which integrates the concept of test-time adaptation into the meta-learning framework and makes it easier for the counting model to adapt to the unknown test distributions. To facilitate the reliable supervision signal at the pixel level, we introduce uncertainty by inserting the dropout layer into the counting model. The uncertainty is then used to generate valuable pseudo labels, serving as effective supervisory signals for adapting the model. In the context of meta-learning, one image can be regarded as one task for crowd counting. In each iteration, our approach is a dual-level optimization process. In the inner update, we employ a self-supervised consistency loss function to optimize the model so as to simulate the parameters update process that occurs during the test phase. In the outer update, we authentically update the parameters based on the image with ground truth, improving the model's performance and making the pseudo labels more accurate in the next iteration. At test time, the input image is used for adapting the model before testing the image. In comparison to various supervised learning and domain adaptation methods, our results via extensive experiments on diverse datasets showcase the general adaptive capability of our approach across datasets with varying crowd densities and scales.

Optimizing accuracy and efficiency in real-time people counting with cascaded object detection

Optimizing accuracy and efficiency in real-time people counting with cascaded object detection

Retraction Note: Multipoint infrared laser-based detection and tracking for people counting

Retraction Note: Multipoint infrared laser-based detection and tracking for people counting

Counting dense object of multiple types based on feature enhancement.

Accurately counting the number of dense objects in an image, such as pedestrians or vehicles, is a challenging and practical task. The existing density map regression methods based on CNN are mainly used to count a class of dense objects in a single scene. However, in complex traffic scenes, objects such as vehicles and pedestrians usually exist at the same time, and multiple classes of dense objects need to be counted simultaneously. To solve the above issues, we propose a new multiple types of dense object counting method based on feature enhancement, which can enhance the features of dense counting objects in complex traffic scenes to realize the classification and regression counting of dense vehicles and people. The counting model consists of the regression subnet and the classification subnet. The regression subnet is primarily used to generate two-channel predicted density maps, mainly including the initial feature layer and the feature enhancement layer, in which the feature enhancement layer can enhance the classification features and regression counting features of dense objects in complex traffic scenes. The classification subnet mainly supervises classifying dense vehicles and people into two feature channels to assist the regression counting task of the regression subnets. Our method is compared on VisDrone+ datasets, ApolloScape+ datasets, and UAVDT+ datasets. The experimental results show that the method counts two kinds of dense objects simultaneously and outputs a high-quality two-channel predicted density map. The counting performance is better than the state-of-the-art counting network in dense people and vehicle counting. In future work, we will further improve the feature extraction ability of the model in complex traffic scenes to classify and count a variety of dense objects such as cars, pedestrians, and non-motor vehicles.

People counting using IR-UWB radar sensors and machine learning techniques

This study aims to detect and count people using impulse radio ultra-wideband radar and machine learning algorithms. However, the data quality, difficulty distinguishing human signals from noise and clutter, and instances where human presence is not detected make it challenging to count multiple humans. To overcome these challenges, we apply wavelet transformation to reduce signal size and use simple moving averages to eliminate noise. Next, we create features based on statistical and entropic properties of the signal and apply several classification algorithms, including ANN, Random Forest, KNN, XGBOOST, and multiple linear regression, to predict the number of people present. Our findings reveal that using the ANN classifier with the Daubechies 4 (db4) wavelet provides better results than other classifiers, with an accuracy rate of 99%. Additionally, filtering the data improves accuracy, and labeling the data after extracting essential characteristics significantly improves the model’s accuracy.

NEW ERA OF VISION TO ENVISION USING YOLO

We provide a new approach to crowd counting in this research that makes use of the You Only Look Once (YOLO) algorithm. We show how well YOLO performs in precisely identifying and counting people in congested environments. Our method seeks to overcome the difficulties associated with crowd observation and analysis in real time. Urban planning, public safety, and event management are just a few of the many areas where crowd counting is important and dynamic. Occlusions, size variations, and congested settings are only a few of the challenges that traditional approaches frequently face in providing precise and timely crowd estimates. This study presents a unique method for accurate and efficient crowd counting that leverages the You Only Look Once (YOLO) algorithm. Through the utilization of YOLO's sophisticated object identification capabilities, our suggested approach provides a reliable way to compute crowd sizes in real-time situations. We provide an in-depth examination of the difficulties involved in crowd counting and show, by intensive testing and assessment on many datasets, how successful our YOLO-based method is. Our findings demonstrate the People counting, Image-based crowd estimation, Video analytics, Data preprocessing, Performance evaluation. Keywords— Crowd counting, Object detection, YOLO (You Only Look Once), Deep learning, Computer vision, Convolutional neural networks (CNNs), Image processing, Crowd analysis, Real-time monitoring, Surveillance systems,

Privacy‐preserving WiFi‐based crowd monitoring

AbstractThe process of estimating the number of individuals within a defined area, commonly referred to as people counting, is of paramount importance in the realm of safety, security and crisis management. It serves as a crucial tool for accurately monitoring crowd dynamics and facilitating well‐informed decision‐making during critical situations. In our current study, we place a special emphasis on the utilization of the WiFi fingerprint technique, leveraging probe request messages emitted by smart devices as a proxy for people counting. However, it is essential to recognize the evolving landscape of privacy regulations and the concerted efforts by major smart‐device manufacturers to enhance user privacy, exemplified by the introduction of MAC addresses randomization techniques. In this context, we designed a crowd monitoring solution that exploits Bloom filters for ensuring a formal deniability, aligning with the stringent requirements set forth by regulations like the European GDPR. Our proposed solution not only addresses the essential task of people counting but also incorporates advanced privacy‐preserving mechanisms. Importantly, it seamlessly integrates with trajectory‐based crowd monitoring, offering a comprehensive approach to managing crowds while respecting individual privacy rights.

Passive Infrared Sensor-Based Occupancy Monitoring in Smart Buildings: A Review of Methodologies and Machine Learning Approaches.

Buildings are rapidly becoming more digitized, largely due to developments in the internet of things (IoT). This provides both opportunities and challenges. One of the central challenges in the process of digitizing buildings is the ability to monitor these buildings' status effectively. This monitoring is essential for services that rely on information about the presence and activities of individuals within different areas of these buildings. Occupancy information (including people counting, occupancy detection, location tracking, and activity detection) plays a vital role in the management of smart buildings. In this article, we primarily focus on the use of passive infrared (PIR) sensors for gathering occupancy information. PIR sensors are among the most widely used sensors for this purpose due to their consideration of privacy concerns, cost-effectiveness, and low processing complexity compared to other sensors. Despite numerous literature reviews in the field of occupancy information, there is currently no literature review dedicated to occupancy information derived specifically from PIR sensors. Therefore, this review analyzes articles that specifically explore the application of PIR sensors for obtaining occupancy information. It provides a comprehensive literature review of PIR sensor technology from 2015 to 2023, focusing on applications in people counting, activity detection, and localization (tracking and location). It consolidates findings from articles that have explored and enhanced the capabilities of PIR sensors in these interconnected domains. This review thoroughly examines the application of various techniques, machine learning algorithms, and configurations for PIR sensors in indoor building environments, emphasizing not only the data processing aspects but also their advantages, limitations, and efficacy in producing accurate occupancy information. These developments are crucial for improving building management systems in terms of energy efficiency, security, and user comfort, among other operational aspects. The article seeks to offer a thorough analysis of the present state and potential future advancements of PIR sensor technology in efficiently monitoring and understanding occupancy information by classifying and analyzing improvements in these domains.

A low-cost automatic people-counting system at bus stops using Wi-Fi probe requests and deep learning

AbstractCounting people is an important part of people-centric applications, and the increase in the number of IoT devices has allowed the collection of huge amounts of data to facilitate people counting. The present study seeks to provide a novel, low-cost, automatic people-counting system for the use at bus stops, featuring a sniffing device that can capture Wi-Fi probe requests, and overcoming the problem of Media Access Control (MAC) randomization using deep learning. To make manual data collection considerably easier, a “People Counter” app was designed to collect ground truth data in order to train the model with higher accuracy. A user-friendly, operating system-independent dashboard was created to display the most relevant metrics. A two-step methodological approach was followed comprising device choice and data collection; data analysis and algorithm development. For the data analysis, three different approaches were tested, and among these a deep-learning approach using Convolutional Recurrent Neural Network (CRNN) with Long Short-term Memory (LSTM) architecture produced the best results. The optimal deep learning model predicted the number of people at the stop with a mean absolute error of ~ 1.2 persons, which can be considered a good preliminary result, considering that the experiment was done in a very complex open environment. People-counting systems at bus stops can support better bus scheduling, improve the boarding and alighting time of passengers, and aid the planning of integrated multi-modal transport system networks.

Multi-scale fusion public gathering recognition based on residual network

With the rapid economic development and high concentration of urban population, people’s income level and quality of life continue to improve, resulting in more and more crowded scenes caused by people going out. Especially in urban commercial centers, transportation hubs, sports venues during important events, tourist attractions, etc., crowd gatherings occur frequently. However, accidents involving crowd gatherings in public places occur frequently, causing heavy casualties and property losses. Therefore, for crowd recognition, this paper proposes a new method to accurately estimate the number of dense crowds. In this method, a density map with accurate pedestrian locations is first generated using the focal inverse distance transform and used as ground truth labels for network training. Then, a multi-scale feature fusion algorithm based on residual network is designed, combining spatial and channel attention mechanisms to improve the accuracy and stability of crowd density estimation. In dense crowds, the phenomenon of overlapping and occlusion of people is very common and serious, making it difficult for existing pedestrian detection methods to distinguish each individual and accurately count the flow of people. To solve this problem, this paper proposes a density map-based method that uses a local maximum detection strategy and a K-nearest neighbor algorithm to convert the density map into the corresponding dense head bounding box. This method can effectively reduce the impact of occlusion and improve the accuracy of people counting. In order to further improve the estimation accuracy, a pattern recognition density peak clustering algorithm is introduced to study the clustered crowds. By treating the head bounding box as an element point, the distance between each element point is calculated, and the density of each point is calculated. Then perform clustering to find the cluster center with the highest density in each class. Finally, by comparing the density of each cluster center with the corresponding density threshold and adopting the corresponding decision-making method, the accuracy of people counting is further improved.

Hybrid optimized temporal convolutional networks with long short-term memory for heart disease prediction with deep features

A heart attack is intended as top prevalent among all ruinous ailments. Day by day, the number of affected people count is increasing globally. The medical field is struggling to detect heart disease in the initial step. Early prediction can help patients to save their life. Thus, this paper implements a novel heart disease prediction model with the help of a hybrid deep learning strategy. The developed framework consists of various steps like (i) Data collection, (ii) Deep feature extraction, and (iii) Disease prediction. Initially, the standard medical data from various patients are acquired from the clinical standard datasets. Here, a One-Dimensional Convolutional Neural Network (1DCNN) is utilized for extracting the deep features from the acquired medical data to minimize the number of redundant data from the gathered large-scale data. The acquired deep features are directly fed to the Hybrid Optimized Deep Classifier (HODC) with the integration of Temporal Convolutional Networks (TCN) with Long Short-Term Memory (LSTM), where the parameters in both classifiers are optimized using the newly suggested Enhanced Forensic-Based Investigation (EFBI) inspired meta-optimization algorithm. Throughout the result analysis, the accuracy and precision rate of the offered approach is 98.67% and 99.48%. The evaluation outcomes show that the recommended system outperforms the extant systems in terms of performance metrics examination.

Enhancing real human detection and people counting using YOLOv8

The ability to accurately recognize and count persons is crucial in many real-world applications, including surveillance, security, and crowd management, making it one of computer vision’s most fundamental tasks. You Only Look Once (YOLO) is one of the most effective deep learning models for object identification and counting in recent years. This research seeks to learn more about the YOLOv8 algorithm for precisely counting people in still photos and moving videos. The YOLO method has been at the forefront of computer vision due to its ability to recognize things in real time. People in a crowd typically overlap and block one other, and perspective effects can result in enormous changes in human size, shape, and appearance in the image, all of which make accurate headcounts challenging.The YOLO methodology and its adaptation for population census are the subject of this research. Results from experiments support the usefulness of the proposed approach. Surveillance, crowd control, traffic monitoring, retail analytics, event management, and urban planning are just some of the potential uses highlighted by the findings of this study. Mean Average Precision (MAP) numbers demonstrate that the identification procedure was successful, and the counting process was accurate to within 100%.

Online Reviews: A Proposal towards Bilingual Approach

In light of the growth of internet usage and its important role in the field of e-commerce, electronic word-of-mouth (eWOM) has been changing people’s behavior and decisions. People count on other users’ opinions and information; they sometimes even make offline decisions based on information acquired online. The purpose of this paper is to conceptualise Bilingual Approach towards eWOM activity from an input-process-output (IPO) perspective; propose a classification framework based on the identified academic literature; analyze eWOM literature in terms of quantitative development and qualitative issues that are useful to both academics and researchers; and provide directions and guidelines for future research studies in eWOM.

Building Ventilation Optimization Through Occupant-Centered Computer Vision Analysis

Abstract Buildings consume about 40 % of all energy. Ventilation plays a significant role in both the energy consumption of buildings and the comfort of occupants. To achieve energy efficiency and comfort, smarter ventilation control algorithms can be employed, such as those with feedback based on CO2 levels. Furthermore, by knowing the current number of people in a space, ventilation can theoretically be adjusted to maintain a constant CO2 level without wasting energy when people are not present. An additional benefit of such control could arise due to occupants’ habits. For example, if a person senses elevated CO2 levels, even if the ventilation system has started operating more intense, they might choose to open a window, potentially compromising energy efficiency. Therefore, if the control algorithm were to maintain a constant CO2 level, occupants may be less likely to open windows. In our work, we explore a model in combination with a custom monitoring system based on computer vision to implement such control. The monitoring system combines outside and inside CO2 sensors with precise people counting based on computer vision to provide data to the model. The model relies on the mass balance equation for CO2 and considers the historical data of the number of occupants and their activities to estimate the overall CO2 generation in indoor spaces. The results suggest that the model can effectively forecast CO2 dynamics with an absolute deviation of 40 ppm. However, it was observed that the analysis of the actual air exchange level could be compromised by several factors.

Unsettling Human (In)Security: Challenging Neglect and Making the Unapparent Appear

The concept of human security, introduced to the global public policy lexicon in 1994, did two things. First, it challenged conventional and dominant state-centric approaches to security to rethink security in universal people-centred terms. Second, in its focus on human life and dignity, it called for a profound transition in security thinking to make all people everywhere matter. This article argues that the development of human security within the UN system has not sufficiently met these objectives. Through an analysis of three key human security sectors— food security, health security and disaster events—it reveals how conventional approaches to human security can be implicated in neglect, invisibility and marginalisation such that those already facing insecurity are made worse off still. A critical human security lens demonstrates that foregrounding recognition, solidarity and agency can help to expose invisibilities and encourage strategies to make people count.