Abstract—In this paper, we introduce an affordable IoT-based solution aiming to increase COVID-19 indoor safety, covering several relevant aspects: 1) contactless temperature sensing 2) mask detection 3) social distancing check. Contactless temperature sensing subsystem relies on Arduino Uno using infrared sensor or thermal camera, while mask detection and social distancing check are performed by leveraging computer vision techniques on camera-equipped Raspberry Pi.
INTRODUCTION Since the last days of the previous year, the occurrence of novel infectious flu-alike respiratory disease COVID-19 caused by SARS-Cov-2 virus (also known as coronavirus) has affected almost every aspect of people’s lives globally. First, it was discovered in China, but spread quickly to other continents in just few weeks. According to [1], until July 11th, 2020, the total number of identified cases was 12,653,451, while taking 563,517 lives worldwide. Common symptoms of coronavirus disease include fever, tiredness, sore throat, nasal congestion [2], loss of taste and smell [3]. In most cases, it is transmitted directly (person to person) through respiratory droplets, but also indirectly via surfaces [4, 5]. Incubation period could be quite long and varies (between 14 and 27 days in extreme cases) [6, 7]. Furthermore, even asymptomatic persons (almost 45% of cases) can spread the disease [7] making the situation even worse. Therefore, the usage of face masks and sanitizers has shown positive results when it comes to disease spread reduction [8]. However, the crucial problem is the lack of approved vaccine and medication [9]. Due to these facts, many protection and safety measures were taken by governments in order to reduce the disease spread, such as obligatory indoor mask wearing, social distancing, quarantine, self-isolation, limiting citizens’ movement within country boarders and abroad, often together with prohibition and cancellation of huge public events and gatherings [10]. Despite the fact that the pandemic seemed weaker at some points, most of safety regulations are still applied due to unstable situation. From workplace behavior to social relations, sport and entertainment, coronavirus disease Nenad Petrović is with the Faculty of Electronic Engineering, University of Niš, Aleksandra Medvedeva 14, 18000 Niš, Serbia (e-mail: [email protected]). Đorđe Kocić is with the Faculty of Electronic Engineering, University of Niš, Aleksandra Medvedeva 14, 18000 Niš, Serbia (e-mail: [email protected]). poses many changes to our everyday routine, habits and activities. In this paper, cost-effective IoT-based system aiming to help organizations respect the COVID-19 safety rules and guidelines in order to reduce the disease spread is presented. We focus on most common indoor measures - people with high body temperature should stay at home, wearing mask is obligatory and distance between persons should be at least 1.5-2 meters. For the first scenario, Arduino Uno microcontroller1 board with contactless temperature sensor is used, while we rely on Raspberry Pi2 single-board computer equipped with camera making use of computer vision techniques for other two scenarios. We decided to use these devices due to their small size and affordability.
BACKGROUND A. OpenCV Python version of OpenCV [11], open-source computer vision library was used for implementation of mask detection and social distance check algorithms. We decide to use it, as it was approved for usage with older Raspberry Pi devices [12]. Face and body detection algorithms rely on the existing OpenCV implementation of Viola-Jones object detection framework based on Haar feature cascades [13]. It is a machine learning approach where cascade function is trained from a large set of positive and negative images. After that, this function is used to detect objects in new images. OpenCV comes with both trainer and detector. However, OpenCV offers pre-defined classifiers for detection of commonly used objects, such as human face, whole body, body and face parts (both front and back for some of them). Therefore, in this paper, we leverage the existing classifiers provided by OpenCV library, as they were enough to cover satisfy the needs of the implemented solution. In [12], face detector provided by OpenCV library was used for control of multimedia reproduction systems based on Raspberry Pi devices within museums and cultural heritage sites, showing acceptable performance, even in real-time use cases. B. MQTT In this paper, MQTT (Message Queuing Telemetry Transport)3 was used for machine-to-machine communication between the involved devices - Raspberry Pi, Arduino, Edge servers and smartphones. It is a lightweight, publish-subscribe messaging protocol on top of TCP/IP. MQTT is designed for use cases where small code footprint is desired or network bandwidth is limited, which is suitable for IoT solution leveraging low-power computing devices presented in this paper. Moreover, publish-subscribe messaging mechanism requires a message broker. For that purpose, we use a Node.js MQTT broker implementation within Node-RED4 deployed on a server residing within the Edge. For IoT devices, corresponding MQTT client libraries were used – PubSubClient5 for Arduino, Paho-MQTT6 for Raspberry Pi and Paho Android Service7 for smartphones. The devices measuring body temperature, detecting masks and social distancing send MQTT messages to Edge servers in cases when a person does not satisfy conditions to pass some of the safety check steps. Furthermore, the Edge server processes the received messages and forwards them to corresponding security workers to notify them about breaking of COVID-19 safety rules. Each message is sent in a form of JSON-encoded string. C. Semantic knowledge representation The role of semantic technology is to enable encoding the meaning of data separately from the content itself and related applications, which provides the ability to understand data, exchange its understanding and perform reasoning on top of it [14]. In this case, the formalization of knowledge is done in a form that is understandable for both humans and machines. Within the semantic knowledge bases, the data is represented with respect to ontologies. An ontology consists of classes, individuals, attributes and relations. Class is an abstract group of objects that have same type. Individuals are instances of these classes. Attributes refer to properties of classes. Furthermore, relations define ways in which classes and individuals can be related. For both the ontologies and facts, the RDF standard language is used. It defines (subject, predicate, object) triplets that are persisted within the semantic triple stores. On the other side, SPARQL is used for query execution against the semantic triple stores. The results are retrieved in order to support different reasoning mechanisms to enable the inference of new facts starting from the existing knowledge base. In IoT systems, ontologies are often used to achieve interoperability and integrate data coming from heterogeneous devices and their sensors in order to enable their control in unified way [15]. For example, in [16], we introduced Smart Grid Response Ontology to enable reasoning about the events that occurred within the IoT-based system for smart grid monitoring and generate adequate response in a particular context. In this paper, we adopt similar approach to semantic representation in context of COVID-19 safety monitoring, but extend it with the elements of spatial reasoning.
RELATED WORK There are several existing works that contain some of the elements relevant to the work presented in this paper. However, to the best of our knowledge, there is no such solution covering all these aspects together to achieve this goal while allowing execution on low-cost IoT devices at the same time. In [17], a dataset for masked face recognition is introduced and its application by different algorithms in context of campus and enterprise coronavirus prevention discussed. Moreover, in [18], a high-accuracy method for facial mask detection using semantic segmentation based on fully convolutional networks, gradient descent and binomial cross entropy was presented. However, performance-wise, it is too heavy for low-power IoT devices, such as Raspberry Pi. On the other side, in [19], a state model-based solution for face mask detection relying on Viola-Jones algorithm in context of ATM center security was described. When it comes to temperature sensing, there are several variants of Arduino-based solutions. In [20], Arduino was used for real-time temperature visualization using MATLAB. However, the used sensor does not allow contactless temperature sensing. Moreover, in [21], a similar system incorporating the usage of smartphones for remote temperature monitoring using Arduino Uno was presented. The system architecture presented in this paper is inspired by our previous work on remote smart grid monitoring anomaly, power consumption monitoring and relay protection using IoT devices (smartphones and Arduino Uno) [16] and video surveillance system relying on Raspberry Pi singleboard computers and Edge servers [22]. Our main goal is to provide a comprehensive solution for COVID-19 safety monitoring which relies on IoT devices as much as possible in order to be affordable at the same time.
to smartphone application used by security guards, so they can arrive to make sure that person does not try to enter the building further. Otherwise, if passenger’s temperature is normal, Arduino will send signal to open the door. After that, passengers proceed to next step of checking – mask detection. For this task, computer vision subsystem based on Raspberry Pi single-board computer equipped with camera module version 110 revision 3 was used. In case that passenger does not wear mask or it does not cover nose, security guards will be informed via MQTT message, so they can provide a mask or warn that person to leave. Otherwise, if the person that is being checked wears mask, the door will be opened. Furthermore, once they enter the building, Raspberry Pi devices check whether social distancing is applied properly or not at given locations. In a similar way, MQTT message will be sent to inform the security guards when social distancing is not applied properly in some of the rooms. On the server side, the MQTT broker and semantic triple store are deployed, while message processing, event logging, reasoning and message forwarding are done. Edge servers receive messages, perform their semantic annotation and reasoning to find the right security guard that will be notified. A simple Android mobile application used by security guards receives MQTT messages from server side and visualizes the data about rule violation and location where it occurred within the building. In Fig. 1, an overview of the proposed IoT-based solution that aims to ensure that COVID-19 safety guidelines are applied properly indoors is given.
IMPLEMENTATION A. System overview Our solution consists of the following subsystems: 1) temperature measurement subsystem based on Arduino Uno 2) computer vision subsystem for mask detection and social distancing check based on Raspberry Pi 3) server side 4) smartphone application for security guards. First, all people that try to the enter building have to pass contactless temperature check. For that purpose, we rely on Arduino Uno equipped with infrared thermometer (such as MLX906148 ) or thermal camera sensor (AMG88339 for example). Moreover, it uses ESP8266 WiFi module for communication with Edge servers using MQTT protocol. In case that person has body temperature higher than normal, the door is locked and MQTT message sent to server, containing both the temperature value and location where it was recorded. Server receives this message, parses it and forwards
CONCLUSION AND FUTURE WORK According to the achieved results, the proposed solution is usable for its purpose under certain performance limitations (such as number of processed frames or measurements per second). Moreover, it relies on both open hardware and free software, being definite and desirable advantage for such systems. In future, it is planned to experiment with various deep learning and computer vision frameworks for object detection on Raspberry Pi in order to achieve higher framerate. Moreover, we would like to extend this solution with environment sensing mechanisms for adaptive building air conditioning and ventilation airborne protection in order to reduce the spread of coronavirus indoors [4, 8, 24], especially during summer. Finally, the ultimate goal is to integrate the system presented in this paper with our framework for efficient resource planning during pandemic crisis [25] in order to enable efficient security personnel scheduling and mask allocation, together with risk assessment based on statistics about respecting the safety guidelines and air qualit
Iot based engineering projects
iot based engineering projects
iot based projects for engineering students
iot based major projects
iot based major projects for ece
iot based projects for mechanical engineering
iot based final year engineering projects
iot based projects for final year engineering students
iot based final year projects for electronics engineering
iot based project
iot based best projects
iot based electrical engineering projects
Thank you for contacting Shield Technologies! Please let us know how we can help you.
For more information please do visit mifratech center.
Note : Find the best solution for electronics components and technical projects ideas
keep in touch with our social media links as mentioned below
Mifratech websites : https://www.mifratech.com/public/
Mifratech facebook : https://www.facebook.com/mifratech.lab
mifratech instagram : https://www.instagram.com/mifratech/
mifratech twitter account : https://twitter.com/mifratech