Search

Counting the number of visitors (people counter) is an activity carried out in many retail stores. This information, although very simple and basic, can be used as one of the main bases for making decisions. In the past, PT Matahari Department Store, Tbk (MDS) has calculated the number of visitors manually with an accuracy of around 90%. The purpose of this research is to design an automated people counter system using the existing CCTV video networks in MDS stores. The research methodology used is data collection methods in the form of literature studies and interviews, and systems development methods in the form of prototyping. The system uses Tensorflow to recognize visitors who enter and leave the store. This system is a prototype and is used as a proof of concept for MDS to make decisions regarding the application of appropriate technology in carrying out the people counter process along with the things that need to be prepared if this system is to be implemented thoroughly in all MDS stores in Indonesia.

This article proposes a convolutional neural network (CNN)-based people counter that classifies a given frame cube to a specific event that indicates people entering or exiting a target area to measure instantaneous people count. For the training of the proposed CNN, a training input frame cube and its corresponding class label that represents a specific event are generated using the proposed counting rules. For mitigating the overfitting problem that may occur in the training of the proposed CNN, data augmentation, and postclass correction using foreground distribution with event probabilities are applied. The experimental results indicate that the proposed method improved the F1 score and accuracy for the cumulative people counting results by up to 9.0% and 14.8%, respectively, compared with those of the benchmark methods, even though it calculated the cumulative count by summing instantaneous people counts, while the benchmark methods were optimized for the calculation of the cumulative count.

CCTV cameras have an important function in the field of public service, especially for convenience. The objects recorded through CCTV cameras are processed into information to support service satisfaction in the community. This study uses the function of CCTV for people counting from objects recorded by a camera. Currently, the process of detecting and tracking people takes a long time to detect all frames. In this study, the frame selection into keyframes uses the mutual information entropy method. The keyframes processing uses the Histogram of Oriented Gradient (HOG) and Kalman filter methods. The proposed method results F1 value of 0.85, recall of 76 %, and precision of 97 % with winStride parameter (12,12), scale 1.05, and the distance of the human object to CCTV 4 meters.

Video or image-based people counting in real-time has multiple applications in intelligent transportation, density estimation or class management, and so on. This problem is usually carried out by detecting people using conventional detectors. However, this approach can be failed when people stay in various postures or are occluded by each other. In this paper, we notice that even a main part of human body is occluded, their face and head are still observable. We then propose a method that counts people based on face and head detection and pairing. Instead of deploying only face or head detector, we apply both detectors as in many cases the human does not turn his/her face to camera then head detector takes advantage. Otherwise, face detector produces reliable results. The fact of combining both head and face detection results will lead to duplicated responses for one person. We then propose a simple yet effective alignment technique to pair a face with a head of a person. Subsequently, the remaining heads and faces which are not paired with any other faces or heads will be added to our people counter to increase the true positive rate. We evaluate our proposed method on four datasets (Hollywood, Casablanca, Wider Face, and our own dataset). The experimental results show an improvement of average precision and recall comparing to the original head or face detectors.

We present a groundtruthing approach which is applicable to large video datasets collected for studying people's behavior, and which are recorded at a low frame per second (fps) rate. Groundtruthing a large dataset manually is a time consuming task and is prone to errors. The proposed approach is semi-automated (using a combination of deepnet and traditional image analysis) to minimize human labeler's interaction with the video frames. The framework employs mask-rcnn as a people counter followed by human assisted semi-automated tests to correct the wrong labels. Subsequently, a bounding box extraction algorithm is used which is fully automated for frames with a single person and semi-automated for frames with two or more people. We also propose a methodology for anomaly detection i.e., collapse on table or floor. Behavior recognition is performed by using a fine-tuned alexnet convolutional neural network. The people detection and behavior analysis components of the framework are primarily designed to help reduce human labor in ground-truthing so that minimal human involvement is required. They are not meant to be employed as fully automated state-of-the-art systems. The proposed approach is validated on a new dataset presented in this paper, containing human activity in an indoor office environment and recorded at 1 fps as well as an indoor video sequence recorded at 15 fps. Experimental results show a significant reduction in human labor involved in the process of ground-truthing i.e., the number of potential clicks for office dataset was reduced by 99.2% and for the additional test video by 99.7%.

With the continuously increasing number of positive cases of COVID-19 in the country, government authorities were left with no alternative but to enforce strict and stringent protective measures to conform to the "new normal". People counter system is the most used vison-based measurement system especially in monitoring hourly footfalls and peak times of customers throughout the day. This type of measurement system is very crucial especially when it comes to crowd control. The general objective of this study is to develop a system that monitors the number of people entering and leaving an establishment via image processing/object tracking using YOLO v3 algorithm to make sure that the maximum capacity of people allowed inside an establishment according to IATF and DTI standards for social distancing purposes is followed and observed.

This paper proposes a portable system to count the number of people entering and exiting a space. The system is developed for use in locations which have only one entrance/exit. The system detects people who walk through the door, calculates the number people, and sends the data to the Blynk and Line Notify servers which forwards the information to a user’s smartphone. To make the system standalone and reduce WiFi connection problems, the GSM cellular data network has been selected to communicate with the Blynk and Line Notify servers. The average accuracy of counting people using the proposed system, is approximately 97.33%.

Detecting and tracking people is a challenging task in a persistent crowded environment as retail, airport or station, for human behaviour analysis of security purposes. Especially during the global spread of SARS-CoV-2 virus that has become part of everyday life in every country, it is important to be able to manage the flows inside and outside buildings indoors. This article introduces an approach to detect and count people when they cross a virtual line. The methods used are based on deep learning and in particular on convolutional neural networks, specifically MobileNetV3 which is used for the detection task and MOSSE filter which is used for the tracking phase. The hardware system assembled for people counting is inexpensive, as it is formed by Raspberry Pi4 and a Picamera module v2. These devices have already been installed in some supermarkets and museums in the center of Italy, precisely in the area of the Marche region.

The spread of mobile phones made it easy to estimate person-flow for corporate marketing, crowd analysis, and countermeasures for disaster and disease. However, due to recent privacy concerns, regulations have been tightened around the world and most smartphone operating systems have increased privacy protection. To solve this, in this study, we propose the person-flow estimation technique with preserving privacy. We use 3D People Counter which can record only the time and direction of passing people, a person’s height, and walking speed, therefore it preserves privacy from the moment of collecting data. To estimate people’s in-out data, we propose four methods and they use some of the sensor data above in different combinations. We compared these methods and the height-based method could estimate about 79% of the sensor data as in-out data. Additionally, we also created a system to interpolate in-out data into person-flow data and to visualize it. By using this method, we believe that it can be used for the purposes described in the beginning.

RESUMEN: A día de hoy, uno de los campos de mayor importancia en la Industria 4.0 es el Internetof-Things (IoT)(Internet de las cosas), con el fin de solucionar pequeñas tareas al ser humano o monitorizar ciertos parámetros. TST Sistemas colaboró en 2018 con Vodafone para el desarrollo de un contador de personas con la finalidad de establecer mapas de calor en un recinto. A raíz de la pandemia mundial, este tipo de tecnologías se ha visto altamente demandada, por lo que se precisa de mejorar el producto y dotar de inteligencia un hardware que muestra datos en bruto. A través de técnicas de Machine Learning, se desea utilizar algoritmos que puedan aportar información útil sobre los usuarios alrededor. En este trabajo se recoge el estudio desde el desarrollo de la idea, pasando por la toma de datos, hasta el desarrollo de los algoritmos. El objetivo principal del sistema a desarrollar es identificar personas dentro o fuera de un edificio y poder localizarlos en el interior de este para el control de aforos. ABSTRACT: Nowadays, one of the most important fields in Industry 4.0 is the Internet-of-Things (IoT), in order to solve small human tasks or monitor certain parameters. TST Sistemas collaborated in 2018 with Vodafone for the development of a people counter in order to establish heat maps in a venue. As a result of the global pandemic, this type of technology has been in high demand, so it is necessary to improve the product and provide intelligence to a hardware that displays raw data. Through Machine Learning techniques, we want to use algorithms that can provide useful information about the users around them. This work includes the study from the development of the idea, through data collection, to the development of algorithms. The main objective of the system to implement is to identify people inside or outside a building and to be able to locate them inside it for capacity control. Máster en Ingeniería de Telecomunicación

Governments have imposed social distancing regulations to counter the coronavirus pandemic. An automated occupancy control system to provide a more cost-effective and efficient way to abide with these safety regulations. With the generalization of the use of digital images, the analysis of movement in video sequences has proved to be an essential tool for various applications such as video surveillance, robotics etc. The advance in video processing algorithms and the fast computational capability, give a possibility to use a video tracking and counting people in real time. Estimating the number of people in real time is useful information for several applications such as security and health management. With the COVID-19 pandemic, the counting of people present in a region of interest is important to control the area occupancy in order to minimize human virus transmission. In this paper we present a finished solution for counting people present in the same area. The system is based on Raspberry Pi and a common camera. The method called BLOB (Binary Large Object) analysis is used. The performance of the system, achieving an average count rate between 95% and 98%.

As evident from an in-depth analysis of the state of the art concerning device tracking through Wi-Fi probes and MAC addresses, these techniques represent an increasingly relevant privacy threat. In this paper we provide design and implementation details of a low-cost and low-power people counter based on the Espressif ESP8266 board, and we explicitly analyze the overall cost of the introduced solution. The proposed device can gather MAC addresses from Wi-Fi packets and is designed to circumvent MAC address randomization, as we demonstrate through practical experiments. Our study also shows that, as IoT devices and components are less and less expensive, even a single person could set up a personal people counting systems to be maliciously installed in urban areas or indoor environments.

Detecting people in video streams involves privacy concerns since people are often unaware of being recorded. Furthermore, cameras and commodity depth sensors are prone to variances of illumination. This work proposes light detection and ranging as sensing technology for a robust and privacy preserving people detection. A people counter is implemented, which captures a coarse human shape by concatenating multiple two dimensional range scans while people pass the sensor’s view. The system works under dynamic illuminated conditions and is running as an embedded node on a single core 700 MHz processing unit. An evaluation with 20 different people crossing an entrance 100 times in each direction was carried out under controlled conditions. Results show a high precision and recall of 100 percent and 99 percent respectively for counting people and determining the walking direction. Privacy is preserved since objects are represented by distance values with a low resolution of 16x16 data points.

Emergence of Industry 4.0 in current economic trend promotes the usage of Internet of Things (IoT) in product development. Counting people on streets or at entrances of places is indeed beneficial for security, tracking and marketing purposes. The usage of cameras or closed-circuit television (CCTV) for surveillance purposes has emerged the need of tools for the digital imagery content analysis to improve the system. The purpose of this project is to design a cloud-based people counter using Raspberry Pi embedded system and send the received data to ThingSpeak, IoT platform. The initial stage of the project is simulation and coding development using OpenCV and Python. For the hardware development, a Pi camera is used to capture the video footage and monitor the people movement. Raspberry Pi acts as the microcontroller for the system and process the video to perform people counting. Experiment have been conducted to measure the performance of the system in the actual environment, people counting on saved video footage and visualized the data on ThingSpeak platform.

[ES] Se desarrolla un dispositivo sensor inalámbrico, capaz de detectar la dirección y sentido del tráfico en la entrada de una instalación, almacenando y enviando los parámetros recolectados mediante el uso de una red LPWAN. Posteriormente estos parámetros podrán ser tratados y analizados mediante una interfaz de usuario, y poder prevenir un aforo superior al recomendado., [EN] A wireless sensor device is developed, capable of detecting the direction and way of traffic at the entrance of a facility, storing and sending the collected parameters through the use of an LPWAN network. Later on, these parameters can be treated and analysed through a user interface, to prevent a number higher than recommended., [CA] Es desenvolupa un dispositiu sensor sense fils, capaç de detectar la direcció i sentit del trànsit en l'entrada d'una instal·lació, emmagatzemant i enviant els paràmetres recol·lectats mitjançant l'ús d'una xarxa LPWAN. Posteriorment aquests paràmetres podran ser tractats i analitzats mitjançant una interfície d'usuari, i poder previndre un aforament superior al recomanat. En concret, s'estudien les diferents tecnologies de comunicació sense fils de baix cost, entrant detalladament de les característiques i la composició d'aquestes xarxes, per a poder determinar la mes indicada per al desenvolupament del projecte. A més, es dissenya la xarxa i l'aplicació web, per a visualitzar les dades de manera efectiva i pràctica.

To learn a reliable people counter from crowd images, head center annotations are normally required. Annotating head centers is however a laborious and tedious process in dense crowds. In this paper, we present an active learning framework which enables accurate crowd counting with limited supervision: given a small labeling budget, instead of randomly selecting images to annotate, we first introduce an active labeling strategy to annotate the most informative images in the dataset and learn the counting model upon them. The process is repeated such that in every cycle we select the samples that are diverse in crowd density and dissimilar to previous selections. In the last cycle when the labeling budget is met, the large amount of unlabeled data are also utilized: a distribution classifier is introduced to align the labeled data with unlabeled data; furthermore, we propose to mix up the distribution labels and latent representations of data in the network to particularly improve the distribution alignment in-between training samples. We follow the popular density estimation pipeline for crowd counting. Extensive experiments are conducted on standard benchmarks i.e. ShanghaiTech, UCF_CC_50, MAll, TRANCOS, and DCC. By annotating limited number of images (e.g. 10% of the dataset), our method reaches levels of performance not far from the state of the art which utilize full annotations of the dataset.

Object detection is gaining popularity in a spectrum of domains such as pose detection, self-driving cars, optical character recognition among others. However, it has scarcely been explored in retail settings. Integrating object detection with data science techniques, mainly quantitative statistics, can help derive valuable insights about consumer patterns. The main aim of this paper is to detect people from a video feed and store it in a database that can be queried using multiple aggregations to estimate several metrics. In the context of shops or malls, novel statistical measures such as footfall, conversion rate, and heat maps among others can be obtained. This will help in making data driven decisions based on historical data and patterns. Estimating the busy hours of the store can help the owners to optimize the staff allocation. The footfall trends can be used to evaluate the effectiveness of the marketing campaigns. In this paper, the proposed method is able to detect and count people from CCTV video feeds with an accuracy of 71.4% using a model suitable for devices with low computational power such as Raspberry Pi. The collected data is used to plot the footfall versus time graph.

Home Automation is a new technology that is growing in this era. It has the capability to provide supporting systems to the fast changing world, elderly and physically challenged people. This paper explains the use of overall design which has been built and implemented. The Wireless Home Automation System (WHAS) with the help of Voice Recognition Application, makes the installation and implementation, cheaper and easy to use. Foot Step Counter is the second part of this paper which deals with the automatic switching of lights. Its need is, when there is no time to start the application, or to connect through Blue-tooth, the person passes door and lights within the room will be switched. In this paper, we have observed range of voice recognition application to be 1.5 meters. The voice recognize accuracy achieved is 85.25% with four parameters of access groups.

Diplomski rad teme Brojač osoba zasnovan na kameri ima za zadatak razviti i ispitati programsko rješenje brojanja prolaznika u video zapisu pomoću OpenCV biblioteka, C++ programskog jezika i Microsoft Visual Studio programskog okruženja. Za potrebe rada razvijena su dva programska rješenja zasnovana na funkcijama za računanje apsolutne razlike između dva okvira i ukupnog optičkog toka okvira i testirana na pet video zapisa s različitim uvjetima. Analizom rezultata uočene su prednosti i razlike programskih rješenja u danim uvjetima. Budući da postoje brojne verzije brojača osoba zasnovane na različitim tehnologijama, cilj rada je informirati buduće korisnike o ponašanju određene verzije u određenim uvjetima kako bi se skratio put istraživanja pri implementaciji. The thesis named Camera-based people counter has the task of developing and examining two software solutions for counting using the OpenCV library, the C ++ programming language, and the Microsoft Visual Studio programming environment. Two software solutions based on the functions cv :: absdiff and cv :: calcOpticalFlowFarneback were developed for the purpose of the thesis and tested on five videos with different conditions. By analyzing the results, the advantages and disadvantages of the program solutions in the given conditions were observed. Since there are numerous versions of people counters based on different technologies, the purpose of the thesis is to inform future users about the behavior of a specific version under certain conditions to shorten the implementation path.

Real-Time Campus University Bus Tracking Mobile Application is a mobile application to help campus members detect the current location of the bus in real-time. Real-Time Campus University Bus Tracking Mobile Application is a hybrid mobile application. However, for this development, it is developed for Android user only. It can show updated estimation time arrival and the number of persons inside the bus. This project using two devices embedded inside the bus, which are GPS Tracker device and IoT people counter device. All devices will transmit the data into cloud database which is Firebase. Real-Time Campus University Bus Tracking Mobile Application is developed as a platform for user to receive the data transmitted from database. Other than that, Student will know the time arrival of the bus and the current quantity of people inside the bus to lead them avoid wasted time knowing that they wait for the bus that pack of passenger. The student also able to make complaint and feedback via the platform. Furthermore, this project using PhoneGap as a tool platform to develop the application. The GPS Tracker device using Arduino and IoT people counter using Raspberry PI to transmit data. Overall this project using the reusability techniques and Agile method to complete all the system which it is involved four interactions to make it full system work as expected.

Currently there is a boom in applications focused on the Internet of Things (IoT), especially in more developed countries. An example of this technological process is the intelligent classrooms, where software and hardware come together to establish wireless communications that allow remote control of various variables. Following this model, this article proposes a solution to improve the public transport service of the city of Loja through a person counter. This counter would work with an algorithm that has been used in a static environment similar to that of a bus with passengers. In addition, processes of said algorithm will be simulated with the MATLAB software in order to verify its accuracy.

Reliable people counting is crucial to many urban applications. However, most existing people counting systems are sensor-based and can only work in some fixed gateways or checkpoints where sensors have been installed. This high dependence on the exact locations of sensors leads to low accuracy. To overcome these limitations, in this paper, we propose a smartphone-based people counting system, Wi-Counter, by leveraging the pervasive Wi-Fi infrastructure. To collect comprehensive Wi-Fi signals and people count information based on crowdsource, Wi-Counter first adopts a preprocessor to overcome the noisy, discrepant, and fragile data based on the Wiener filter and Newton interpolation. It then makes use of the designated five-layer neural network to learn the relation model between the Wi-Fi signals and the number of people. By analyzing the received Wi-Fi signals, Wi-Counter can estimate the number of people based on the resulting model. We have conducted experiments by implementing a prototype of Wi-counter based on smartphones and evaluated the system in terms of accuracy and power consumption in an indoor testbed covering an area of 96 m $^2$ . Wi-Counter achieved a counting accuracy of up to 93% and exhibited reliable and robust performance resisting temporal environmental changes with negligible power usage.

In this paper, we propose an automatic people counting system by using our low-cost Internet-of-Things (IoT) platform consisting of a single camera and Raspberry Pi. In this system, we count the number of moving people in bidirection by observing from a side view. Because the system can determine the height of the people, our system can be used to classify them into adults or children. This system is applied for no people overlapping problem in indoor environment only. The background subtraction and morphological operations are used to extract foreground objects from background images. The experimental results show proposed method can achieve 98% of people counting accuracy. It can also achieve 91% accuracy in adult/child classification. Although the algorithms for the people counting and classification are not novel, our technical contribution is that we have implemented them onto our IoT platform, whose cost is less than 100 US dollars. In addition, the images do not need be sent to the server, but all the image processing is done inside the device and only the results are uploaded to the server. This system can be applied to for customer behavior analysis or security.

This paper presents an application for counting people through a single fixed camera. This system performs the count distinction between input and output of people moving through the supervised area. The counter requires two steps: detection and tracking. The detection is based on finding people's heads through preprocessed image correlation with several circular patterns. Tracking is made through the application of a Kalman filter to determine the trajectory of the candidates. Finally, the system updates the counters based on the direction of the trajectories. Different tests using a set of real video sequences taken from different indoor areas give results ranging between 87% and 98% accuracies depending on the volume of flow of people crossing the counting zone. Problematic situations, such as occlusions, people grouped in different ways, scene luminance changes, etc., were used to validate the performance of the system.

The work in this paper focuses on an implementation of OpenCV in an embedded system like raspberry Pi to create a mini-standalone station for counting people. The key feature of AU-PiCC (Assumption University's raspberry Pi Customer Counter) is to count a number of interested people on target product in a pre-defined area along with a simple face identification to avoid counting duplicates. The experimental results show that this raspberry Pi-based system can be used as a simple people counter station.

El proyecto (PePiCo:People Pi Counter) tiene como propósito implementar un algoritmo capaz de determinar el número de personas que atraviesan una barrera virtual, distinguiendo el sentido de la marcha, para su ejecución sobre un dispositivo Raspberry Pi.

We have developed some human sensing systems that can be used in real time environments, such as on railway station platforms and at railroad crossings, to increase safety in busy public places. We describe an application for counting the number of people remaining in divided areas in large spaces. For this purpose, we installed stereo cameras on ceilings that point downward towards the boundaries of specified areas. The system can track all the people passing in each camera's view and can count the number of people remaining in each area. The system can work robustly even under extreme lighting variations by using the stereo cameras. It was used at “EXPO 2005 Aichi, Japan,” and was active opening hours a day for six months. We have stored all the tracking and counting data to make a precise analysis of the human behavior.

Human Tracking in Computer Vision is a very active up-going research area. Previous works analyze this topic by applying algorithms and features extraction in 2D, while 3D tracking is quite an unexplored filed, especially concerning multi–camera systems. Our approach discussed in this paper is focused on the detection and tracking of human postures using multiple RGB–D data together with stereo cameras. We use low–cost devices, such as Microsoft Kinect and a people counter, based on a stereo system. The novelty of our technique concerns the synchronization of multiple devices and the determination of their exterior and relative orientation in space, based on a common world coordinate system. Furthermore, this is used for applying Bundle Adjustment to obtain a unique 3D scene, which is then used as a starting point for the detection and tracking of humans and extract significant metrics from the datasets acquired. In this article, the approaches are described for the determination of the exterior and absolute orientation. Subsequently, it is shown how a common point cloud is formed. Finally, some results for object detection and tracking, based on 3D point clouds, are presented.

Accurate visitor counting is needed in the research of people flows and people distributions. A case study about the performance of four commercial direction-sensitive over-head camera sensors (Axis and Biodata video camera counters, Sick time-of-flight camera, and Irisys thermographic people counter) in detecting patterns of two and three persons was carried out. Although all the tested sensors except Axis generally performed well with the patterns of two people, error rates of even over 20% were observed with the three-person patterns. Simultaneous false positive and false negative detections also occasionally speciously improved the counting results of Axis and Sick. Due to these flaws the accuracies of the sensors do not allow reliable people flow modeling or estimation of room occupancy levels, but is sufficient e.g. for measuring the order of magnitude of customer flows. However, as there are differences between separate counting sites, the results are only indicative in general.

This article introduces a multi-camera system for real-time people counting. The proposed system is built from INT3-Horus, a framework for intelligent monitoring and activity interpretation. The system uses an indoor overhead video camera and a lateral view video camera to detect people moving freely in smart spaces. The segmentation is performed from both synchronized input videos. Then, information is fused to enhance the overall efficiency. The people counting system is flexible in detecting individuals as well as groups. Also, people counting is independent of the trajectories and possible occlusions of the humans present in the smart space. The initial results offered are very promising.

The ability to count people from video is a challenging problem. The scientific challenge arises from the fact that although the task is relatively well-defined, the imaging scenario is not well constrained. The background scene can be uncontrolled along with the illumination being complex and varying. Additionally, the spatial and temporal image resolution is usually poor. The context of most works in people counting is in counting pedestrians from single frames in outdoor settings or moving subjects in indoor settings from standard frame rate video. There is little work done on counting of persons in varying poses, who are mostly static (sitting, lying down), in very low frame rate video (4 frames per minute), and under harsh illumination variations. In this chapter, we explore a design that handles illumination issues at the pixel level using photometry-based normalization, and pose and low-movement issues at feature level by exploiting the spatio-temporal coherence that is present among small body part movements. The motion of each body part, such as the hands or the head, will be present even in mostly static poses. These short duration motions will occur spatially close together over the image location occupied by the subject. We accumulate these using a spatio-temporal autoregressive (AR) model to arrive at blob representations that are further grouped into people counts. We show quantitative performance on real datasets.

This paper presents the design principles underlying the video nodes of long-lifetime wireless networks. The hardware and firmware architectures of the system are described in detail, along with the system-power-consumption model. A prototype is introduced to validate the proposed approach. The system mounts a Flash-based field-programmable gate array and a high-dynamic-range complementary metal-oxide-semiconductor custom vision sensor. Accurate power measurements show that the overall consumption is 4.2 mW at 3.3 V in the worst case, thus achieving an improvement of two orders of magnitude with respect to video nodes for similar applications recently proposed in the literature. Powered with a 2200-mAh 3.3-V battery, the system will exhibit a typical lifetime of about three months. © 2011 IEEE.

Wireless Camera Network (WCN) nodes differ from traditional Wireless Sensor Network (WSN) nodes because of the huge amount of data generated by the sensing element. In order to be able to operate on batteries for a long period, a WCN node needs to extract the information contained into an image and synthesize it into a short message that can be wirelessly transmitted with a limited amount of power. Unfortunately, this approach typically brings the power consumed by the processing unit at the same level as the transceiver, known to be the major source of power consumption in WSN. Thus, there is the need to design efficient algorithms that can be implemented on low-power devices. In this paper we propose the implementation of a Dijkstra-based people counting algorithm for ultra-low-power FPGAs. The developed code has been integrated on the prototype of a WCN node that consumes as little as 5mW.

We describe the implementation in a self-standing systemof a novel contrast-based binary CMOS imaging sensor.This sensor is characterized by very low power consumptionand wide dynamic range, which makes it attractive forwireless camera network applications. In our implementation,the sensor is interfaced with a Flash-based FPGA processor,which handles data readout and image processing.This self-standing camera node is configured as a system forcounting persons walking through a corridor. Simple featuresare extracted from each image in a video stream at 30fps. A classifier is designed based on the temporal evolutionof these features, which is modeled as a Markov chain. Thevideo stream is then segmented into intervals correspondingto individual persons crossing through the field of view. Experimentalresults are shown in cross-validated tests overreal sequences acquired by the camera.

Based on area and color analyses, a cost-effective bi-directional people counter dedicated to the pedestrian flow passing through a gate or a door is proposed. Firstly, the passing people are roughly counted with the area of people projected on an image captured by a zenithal video camera. The moving direction of the pedestrian can be recognized by tracking each people-pattern with an analysis of its HSI histogram. To improve the accuracy of counting, the color vector extracted from the quantized histograms of intensity or hue is introduced to refine the early counting. Besides, the inherent problems of both people touching together and merge/split phenomenon can be overcome. Experimental results show that an 100% accuracy of bi-directional counting can be achieved if the people number of a people-touching pattern is less than six.

This paper deals with an application of image sequence analysis. In particular, it addresses the problem of determining the number of people who get into and out of a train carriage when it's crowded, and background and/or illumination changes. The proposed system analyzes image sequences and processes them using an algorithm based on the use of several morphological tools, which are presented in detail in the paper. Teoría de la Señal y Comunicaciones