Edge AI and IoT-based home security and safety system

Abstract

A system for home security and protection using artificial intelligence. The system includes: a computing unit configured with at least one quad-core processor and at least four gigabytes of RAM, the computing unit comprising an integrated wireless communication module and an Ethernet port; a webcam physically connected to a Universal Serial Bus interface of the computing unit, wherein the webcam is configured to continuously capture live image frames of a monitored access point; an electromagnetic electromechanical door lock that is mechanically mounted on a door structure and electrically coupled to a relay module, the relay module being in electrical connection with multi-purpose input / output pins of the computing unit; at least one fire and smoke detection module comprising a sensor coupled to an analog-to-digital converter, the output of which is electrically connected to the computing unit; an ambient light sensor comprising a light-dependent resistor arranged in a voltage divider circuit and coupled to the computing unit; an infrared occupancy detection sensor that is electrically connected to the computing unit; a piezoelectric buzzer that is electrically connected to the processing unit via a transistor driver circuit; and A regulated power supply with a DC-DC step-down converter configured to supply a direct current of five volts to the processing unit and associated sensors, the processing unit being configured to perform real-time facial recognition via the webcam input, actuate the magnetic door lock via the relay module upon detection of a known face, trigger the piezoelectric buzzer upon a hazard output from the fire and smoke sensor, and switch electrical loads connected via the relay module in response to outputs from the ambient light and infrared presence sensors.

Description

Technical field of the invention

[0001] The present invention relates to the field of intelligent home automation and security systems. In particular, it is an edge AI-powered security device for the home, integrated into IoT communication frameworks. Facial recognition, intrusion detection, and security monitoring are implemented using lightweight machine learning models on embedded hardware platforms such as Raspberry Pi. The invention also includes an electromechanical locking mechanism controlled by AI-driven recognition logic and coupled with fire detection, automatic lighting control, and real-time notification mechanisms. Background of the invention

[0002] With the rapid spread of smart home technology and the Internet of Things (IoT), the demand for autonomous and intelligent systems that ensure security, comfort, and energy efficiency is increasing. Existing security devices, such as conventional CCTV cameras or simple IoT-enabled alarm systems, often lack the necessary intelligence to distinguish between known and unknown individuals. This frequently leads to false alarms and inconvenience for the homeowner.

[0003] Deep learning-based approaches like convolutional neural networks (CNNs) and region-based CNNs offer robust recognition capabilities, but require significant computational resources and large training datasets. These requirements make them unsuitable for real-time edge deployment on resource-constrained embedded systems. Similarly, purely IoT-driven devices without embedded AI modules are limited in their autonomy, as they rely heavily on cloud-based inference engines. This results in latency, dependence on network connectivity, and potential privacy risks.

[0004] While conventional fire and smoke detectors effectively trigger local alarms, they lack intelligent communication systems that can instantly notify the homeowner remotely. Furthermore, most lighting automation systems rely solely on motion detection and do not include adaptive light sensors for energy-efficient use of daylight.

[0005] Therefore, there is a great need for a compact, cost-effective, and edge-deployable security system that combines efficient AI-driven detection, IoT-based communication, and autonomous real-time actuation, while also being energy-efficient and resilient under varying environmental conditions.

[0006] The field of home security has undergone rapid technological development over the past two decades, driven by the proliferation of inexpensive sensors, the development of the Internet of Things (IoT), and the integration of artificial intelligence (AI) into embedded systems. Traditionally, home security relied on mechanical locks, burglar alarms, and CCTV. While these conventional approaches offered basic deterrence, they lacked the ability to intelligently detect and respond to threats in real time. Mechanical locks could be easily broken or duplicated, and traditional alarm systems were limited to audible alerts within the building, offering no communication capabilities to notify the homeowner remotely.CCTV cameras expanded surveillance capabilities by recording video footage, but their passive nature required continuous human monitoring and review of the stored recordings. Such systems were reactive rather than proactive, providing evidence only after an incident had occurred, rather than preventing or mitigating break-ins in real time.

[0007] With the advent of the IoT, a new generation of home security products emerged. These systems used internet connections to transmit alerts, sensor readings, and video streams to the homeowner's smartphone or a central monitoring service. Companies like Ring and Nest popularized smart doorbells and connected cameras, allowing homeowners to remotely view live feeds and receive push notifications of detected motion. While these systems improved the convenience and accessibility of home monitoring, they also introduced significant drawbacks. Most IoT-based security products relied heavily on cloud-based processing, resulting in latency, dependence on an uninterrupted internet connection, and potential privacy risks due to data transmission and storage on remote servers.Furthermore, subscription-based cloud services increased the long-term cost burden for consumers. Another limitation was the limited intelligence of these devices, as they typically used simple motion sensors or image differentiation techniques, which generated a large number of false alarms caused by pets, moving shadows, or environmental disturbances.

[0008] Parallel to the IoT, the field of computer vision made significant strides with the development of deep learning models such as convolutional neural networks (CNNs). These models offered higher accuracy in object and face recognition compared to earlier, feature-based approaches like histogram of oriented gradients (HOG) and support vector machines (SVM). Several scientific studies and commercial implementations demonstrated the robustness of CNNs in handling variations in lighting, face orientation, and occlusion. However, the drawback of CNN-based models was their high computational cost, making them unsuitable for real-time use on resource-constrained devices such as microcontrollers or low-power embedded boards. Their reliance on graphics processing units (GPUs) or cloud-based servers limited their feasibility for edge-based smart home applications.Although deep learning achieved impressive recognition accuracy, its energy consumption, cost, and infrastructure requirements limited its widespread integration into home security systems, particularly in developing regions where cost and connectivity constraints are critical.

[0009] In parallel, lightweight computer vision methods such as Haar cascade classifiers have gained popularity in embedded vision applications because they efficiently detect objects like faces in real-time video streams with limited computing resources. While Haar cascade techniques are not as accurate as their deep learning counterparts, they offer practical advantages for real-time edge deployment, especially on devices like the Raspberry Pi. Several experimental implementations have shown that Haar cascade detection can achieve reasonable accuracy under controlled lighting conditions and process multiple frames per second without GPU acceleration. However, Haar cascade models struggle under challenging conditions such as low light, heavy shadows, or partially obscured faces.Their recognition accuracy is inherently lower than that of deep neural networks, and they are more vulnerable to spoofing attacks, such as those using printed photos or video playback. Nevertheless, their lightweight processing profile has made them viable candidates for edge-based home security prototypes that seek a balance between speed, cost, and autonomy.

[0010] In addition to facial recognition, smart home security systems have also been deployed in the areas of fire detection and energy management. Traditional smoke detectors function as standalone alarm units, emitting local audible signals when smoke is detected. While these devices effectively alert people present in the home, they lack the necessary integration with communication technologies to notify absent homeowners. This limitation can delay emergency response and potentially lead to greater loss of life and property. In response, IoT-enabled smoke and fire detectors have been introduced, capable of sending push notifications or SMS alerts to connected devices. However, these solutions have faced the same drawbacks, such as cloud dependency, vulnerability to network outages, and the security gaps that plague IoT ecosystems.Many low-cost IoT sensors also suffer from poor calibration and limited resilience in detecting subtle environmental changes, leading to both false alarms and delayed detection.

[0011] Another important aspect of smart horns is energy efficiency, particularly in the automation of lighting and HVAC systems. Conventional lighting systems with motion sensors switch the lights on when movement is detected and off again after a defined period of inactivity. While these systems are effective to a certain extent, they don't take into account available natural light and often switch on the lights unnecessarily during the day. More advanced solutions use light sensors to utilize daylight, dynamically adjusting the electric lighting to the natural light level. However, such systems are usually standalone and not tightly integrated with security systems, leading to fragmented smart home ecosystems.Furthermore, poorly designed automation systems can cause inconvenience through abrupt switching or a lack of adaptation to user preferences, which reduces user acceptance despite energy-saving advantages.

[0012] Integrating multiple subsystems—facial recognition, door access control, fire protection, and energy management—into a coherent architecture remains a challenge. Existing commercial solutions often focus on one area, such as security or energy management, but lack comprehensive security and automation capabilities. Homeowners may have to purchase multiple devices from different vendors, leading to compatibility issues, higher costs, and more complex maintenance. Proprietary communication protocols further exacerbate interoperability problems and make it difficult to create a unified, vendor-agnostic smart home ecosystem. Another drawback of many off-the-shelf systems is that they are primarily designed for urban, well-connected areas.In areas with interrupted internet connections or unreliable power supplies, their performance drops drastically, leaving large parts of the world's population underserved.

[0013] Another critical drawback of cloud-based and IoT-heavy solutions is data privacy. The transfer of facial recognition data, video streams, and occupancy information to third-party servers raises significant concerns about surveillance, unauthorized access, and misuse of personal data. High-profile data breaches and growing concerns about digital privacy have deterred consumers from adopting cloud-based home security systems. Edge computing, where processing occurs locally on the device itself, offers an attractive alternative because it minimizes the transfer of sensitive data outside the home. The challenge, however, lies in developing techniques efficient enough to operate on edge devices without compromising detection accuracy or response time.This balance between performance and efficiency is still the subject of active research and development.

[0014] Furthermore, existing solutions are not particularly resistant to attacks and spoofing. Many commercial facial recognition systems are vulnerable to simple hacking attempts, such as presenting a printed photo or a video playback of an authorized user in front of the camera. Without liveness detection or multimodal authentication, these systems offer only superficial security. Advanced liveness detection methods, such as blink detection, head movement detection, or 3D depth sensing, require additional sensors and computing resources, increasing both complexity and cost. Therefore, many smart security products in the lower to mid-price range sacrifice robustness against spoofing for affordability, leaving homeowners vulnerable to sophisticated break-in attempts.

[0015] Another problem with existing security systems is scalability. High-end, AI-powered security solutions with reliable performance are unaffordable for the average household, while low-cost alternatives compromise on accuracy, robustness, and integration. Furthermore, the complexity of installation and configuration can prevent widespread adoption, especially among less tech-savvy users. Systems requiring professional installation and maintenance incur ongoing costs and limit long-term return on investment for homeowners.

[0016] Overall, the current landscape of residential security solutions is characterized by a trade-off between cost, accuracy, computing power, and reliability. Mechanical and conventional alarm systems, while affordable, lack the necessary intelligence. IoT-based devices improve connectivity and remote monitoring but introduce dependencies on cloud services, privacy risks, and increased vulnerability to network outages. Deep-learning-based detection models achieve high accuracy but are impractical for cost-effective, real-time edge deployment. Lightweight image processing techniques enable embedded deployment but reach their limits in harsh environments and are susceptible to spoofing. While fire and energy management subsystems exist, they are often fragmented and poorly integrated into security frameworks.Commercial offerings either focus on only one function or are too expensive and complex for widespread acceptance.

[0017] Therefore, there remains a need for a comprehensive, cost-effective, and edge-deployable home security system that integrates real-time facial recognition, autonomous access control, IoT-based alerting, fire monitoring, and energy-efficient automation into a single device. Such a system must overcome existing limitations regarding accuracy and efficiency, network dependency, privacy risks, and fragmented integration, thereby addressing the shortcomings of current solutions and making smart home security accessible to a wider population. Summary of the invention

[0018] The invention discloses a multifunctional security and protection device for the home, comprising a Raspberry Pi-based processing unit, a USB / webcam for real-time facial recognition with hair cascade classifiers, an electromechanical magnetic door lock for autonomous access control, IoT-based communication modules for alarm transmission, and additional sensors, including fire / smoke detectors and ambient light sensors.

[0019] When a face is detected, the hair cascade model extracts features that are compared to pre-stored numerical vectors of authorized users. If there is a match, the magnetic lock is activated to unlock the door. If faces are not recognized, the system captures an image of the intruder, attaches it to an email notification, and sends it to the homeowner via the SMTP protocol.

[0020] The device also features energy-saving mechanisms where light sensors detect ambient brightness and automatically regulate the interior lighting. Integrated fire and smoke detectors trigger both local alarms and remote IoT alerts, thus increasing the system's security.

[0021] The main objective of the present invention is to provide a comprehensive home security system that seamlessly integrates artificial intelligence and the Internet of Things into a single, network-edge platform. The invention enables real-time decision-making by incorporating lightweight facial recognition and identification technologies that operate efficiently on resource-constrained hardware such as the Raspberry Pi. This eliminates the need for external cloud servers and ensures data privacy, low latency, and continuous availability. A further objective is to ensure that the system not only recognizes authorized individuals and grants access but also issues immediate alerts with photographic evidence when intruders are detected. This enhances the proactive security of residential environments.

[0022] Another important objective of the invention is to provide a device with an electromechanical door locking mechanism that operates autonomously based on the results of the recognition technology and ensures that only verified persons gain access. This eliminates the need for conventional keys or passwords and reduces the security risks associated with physical duplication or password misuse. A further objective of the invention is to provide an intelligent communication module, implemented via SMTP protocols, capable of sending real-time email notifications with image attachments to homeowners. This ensures that unauthorized access attempts are detected immediately, regardless of the homeowner's location.

[0023] Furthermore, the invention goes beyond intrusion detection and contributes to household security and energy efficiency. It integrates fire and smoke detection modules that trigger both local alarms and IoT-based remote alerts, thus enabling an early response to emergencies. In addition, the invention aims to save energy and reduce electricity costs. To this end, light sensors and occupancy-based automation circuits are integrated, regulating lighting and fan operation according to ambient conditions and user presence. The invention is thus intended to offer a multi-dimensional security concept that not only prevents unauthorized access but also improves overall household security and energy sustainability.

[0024] Another objective of the invention is modularity and scalability, enabling the system to be used in various households or small commercial environments without complex installation or excessive costs. By utilizing open-source libraries and cost-effective hardware, the invention aims to make smart home security accessible to a wider population, including those in regions with limited connectivity and budget constraints. A further objective is to ensure the system's resilience, reliability, and adaptability under diverse environmental conditions, while simultaneously allowing for the future integration of advanced detection technologies and additional sensors. This ensures the system remains robust, future-proof, and adaptable to emerging security challenges. BRIEF DESCRIPTION OF THE FIGURE

[0025] These and other features, aspects, and advantages of the present invention will be better understood if the following detailed description is read with reference to the accompanying drawing, in which the same symbols consistently represent the same parts. The following applies: Fig. Figure 1 shows a block diagram of an edge AI and IoT-based security system for private households.

[0026] Experts will also recognize that the elements in the drawing are shown for the sake of simplicity and are not necessarily to scale. For example, the flowcharts illustrate the process by highlighting the main steps to enhance understanding of the aspects of this disclosure. Furthermore, with regard to the design of the device, one or more components of the device may be represented in the drawing by conventional symbols, and the drawing may show only the specific details relevant to understanding the embodiments of this disclosure, so as not to clutter the drawing with details that are readily apparent to those skilled in the art after reading this description. Detailed description of the invention

[0027] For a better understanding of the inventive principles, reference is made below to the embodiment shown in the drawing, which is described in specific terminology. However, this does not limit the scope of the invention. Changes and further modifications of the illustrated system, as well as further applications of the inventive principles, are possible, as would normally occur to a person skilled in the art in the field of invention.

[0028] It is clear to the person skilled in the art that the preceding general description and the following detailed description are exemplary and explanatory of the invention and are not intended as a limitation of it.

[0029] References in this specification to “an aspect”, “another aspect”, or similar expressions mean that a particular feature, structure, or property described in connection with the embodiment is included in at least one embodiment of the present disclosure. Therefore, occurrences of the expressions “in one embodiment”, “in another embodiment”, and similar expressions in this specification may all refer to the same embodiment, but need not.

[0030] The terms "includes," "include," or other variations thereof are intended to cover non-exclusive inclusion, so that a process or method that includes a list of steps may not only contain those steps but may also include other steps not expressly listed or inherent in such process or method. Likewise, the statement "includes..." in the case of one or more devices, subsystems, elements, structures, or components does not, without further limitations, preclude the existence of other devices, subsystems, elements, structures, components, or additional devices, subsystems, elements, structures, or components.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by a person skilled in the art in the field of the invention. The system, methods, and examples provided here serve only for illustration and are not to be construed as a limitation.

[0032] Embodiments of the present disclosure are described in detail below with reference to the attached drawing.

[0033] In Fig.Figure 1 shows a block diagram of a quantum-enhanced system for detecting fake news. The system comprises: a Raspberry Pi 4 computing unit (102) with at least one quad-core processor and at least four gigabytes of RAM, an integrated wireless communication module, and an Ethernet port; a webcam (104) physically connected to a USB interface of the Raspberry Pi and continuously recording live images of a monitored access point; an electromagnetic, electromechanical door lock (106) mechanically mounted to a door structure and electrically connected to a relay module, which is electrically connected to the common input / output pins of the Raspberry Pi; and at least one fire and smoke detection module (108) comprising an MQ-series sensor with an analog-to-digital converter, the output of which is electrically connected to the Raspberry Pi.an ambient light sensor (110) comprising a light-dependent resistor arranged in a voltage divider circuit and connected to the Raspberry Pi via an MCP3008 converter; an infrared presence detection sensor (112) electrically connected to the Raspberry Pi; a piezoelectric buzzer (114) electrically connected to the Raspberry Pi via a transistor driver circuit;and a regulated power supply (116) comprising a step-down DC-DC converter configured to supply five volts direct current to the Raspberry Pi and the associated sensors, the Raspberry Pi executing instructions stored therein to perform real-time facial recognition via the webcam input, to actuate the magnetic door lock via the relay module upon detection of a known face, to trigger the piezoelectric buzzer upon a hazard output from the fire and smoke sensor (116a), and to switch electrical loads connected via the relay module in response to outputs from the ambient light and infrared presence sensors.

[0034] In one embodiment, the magnetic lock (106) comprises a cylindrical piston enclosed in a metal housing. When energized by a 5-volt DC signal supplied via the relay module, the piston is moved into a retracted position. The relay module includes a single-pole, single-stage switching element that is energized via the general-purpose input / output pin of the Raspberry Pi. The magnetic lock is mounted such that, in the de-energized state, the piston engages a strike plate attached to the door frame, thus mechanically securing the door in a locked position.

[0035] In one embodiment, the fire and smoke detection module (116a) comprises a semiconductor sensor MQ-2 mounted on a printed circuit board. The sensor includes a heating element and a tin dioxide-sensitive layer. The sensor is configured to change its resistance in response to the smoke or gas concentration. The analog output of the sensor is passed through the analog-to-digital converter MCP3008 and transmitted to the Raspberry Pi as a digital voltage value. The converter is electrically connected via a serial peripheral interface bus, enabling continuous measurement of the environmental conditions.

[0036] In one embodiment, the ambient light sensor (110) comprises a light-dependent resistor connected in series with a fixed-value resistor to form a voltage divider, wherein the divider is powered by the regulated 5-volt supply and wherein the output node of the divider is electrically connected to a channel of the analog-to-digital converter MCP3008, wherein the digital output is processed by the Raspberry Pi to calculate corresponding illuminance levels, wherein the Raspberry Pi is also coupled to interior lighting loads via a relay driver circuit, so that the loads are automatically switched on when the illuminance falls below a predetermined threshold.

[0037] In one embodiment, the infrared presence sensor (112) comprises a passive infrared sensor module (PIR) housed in a plastic housing with a Fresnel lens.

[0038] The PIR sensor comprises dual pyroelectric sensor elements configured to output a high-level digital signal when infrared radiation patterns change due to human movement. This output signal is routed directly to a digital input pin of the Raspberry Pi, and the Raspberry Pi activates electrical devices via the relay module only when both the PIR sensor output indicates occupancy and the ambient light sensor output indicates insufficient illumination.

[0039] In one embodiment, the regulated power supply (116) comprises a step-down converter with an input terminal connected to a 12-volt DC power source and an output terminal regulated to 5 volts DC. The step-down converter is mounted on a printed circuit board with an inductor, a Schottky diode, and an electrolytic capacitor to provide voltage regulation, the regulated 5-volt output powering the Raspberry Pi, the relay modules, the sensors, and the webcam, thus maintaining stable operation of the system under varying load conditions.

[0040] In one embodiment, the piezoelectric buzzer (114) consists of a piezoelectric ceramic disk in a cylindrical housing. The buzzer is connected in series with a transistor switch, the base of which is connected via a current-limiting resistor to a general-purpose input / output pin of the Raspberry Pi. The Raspberry Pi outputs a digital high signal to saturate the transistor and close the circuit. This triggers an audible alarm when hazardous smoke or fire conditions are detected.

[0041] In one embodiment, the relay module comprises an opto-isolated driver circuit with an optocoupler, a transistor, and a freewheeling diode. The module is configured to receive a low-level control signal from the Raspberry Pi's universal input / output pin and actuate an electromechanical relay coil. The relay coil switches an isolated circuit connected to AC or DC household loads, thus providing safe electrical isolation between the control electronics and high-voltage equipment.

[0042] In one embodiment, the webcam (104) comprises a CMOS (complementary metal-oxide-semiconductor) image sensor module with a fixed focal length lens. The module is integrated into a housing that is mounted next to the door frame. The camera's USB output is directly connected to a USB input of the Raspberry Pi, enabling the continuous acquisition of digital video images for processing within the Raspberry Pi's processing unit.

[0043] In one embodiment, the Raspberry Pi computing unit (102) is housed in a protective enclosure with ventilation slots and mounting devices, the enclosure being mechanically attached to an inner wall next to the door, the electrical wiring of the magnetic lock, relay modules, sensors and power supply terminating in appropriate terminals within the enclosure, thereby providing a structurally integrated and compact embedded system for smart home security applications.

[0044] The present invention describes a security system for private households based on artificial intelligence and IoT, which combines a structural hardware framework with an embedded software pipeline to enable intelligent, autonomous decisions. At the heart of the system is a Raspberry Pi 4 single-board computer, chosen for its quad-core processor, four gigabytes of RAM, an integrated wireless communication module, and an Ethernet interface. The Raspberry Pi serves as the central processing unit and orchestrates the data flow between sensors, actuators, and communication interfaces. A webcam connected via a USB interface is positioned next to the monitored door frame to capture continuous live video streams. These frames are transmitted to the Raspberry Pi in real time for further analysis.

[0045] The technique, running on the Raspberry Pi, comprises two closely linked phases: face detection and face recognition. Detection is achieved by initializing a hair-cascade classifier trained on frontal facial features. Each incoming video frame is preprocessed by converting its RGB pixel structure into a grayscale matrix, significantly reducing computational effort. A sliding windowing process is then applied to the image, extracting hair-like rectangular features representing edges, lines, and intensity variations. The classifier compares these features against pre-trained thresholds. If the threshold conditions are met, the region is flagged as a potential face. The detected facial area is then cropped, normalized for scale and brightness, and passed to the recognition phase.

[0046] During the recognition phase, the system calls a Dlib-based face encoding module. This module computes a 128-dimensional embedding vector that uniquely represents the geometry and features of the recognized face. During system initialization, a database of known individuals is loaded into the Raspberry Pi's memory, with each person represented by several pre-coded vectors derived from sample images. Recognition is performed by comparing the vector of the recognized face with the vectors stored in memory. The comparison is carried out using Euclidean distance metrics, with a distance below 0.6 considered a confirmed match. If a match is confirmed, the Raspberry Pi outputs a control signal via its GPIO pins to activate a relay module, which in turn operates the magnetic lock.The solenoid valve retracts its piston and releases the strike plate mounted on the door frame, thus unlocking the door only for authorized persons.

[0047] If no match is found, or if the recognition score falls below a probability threshold of 85%, the technology triggers the IoT communication subsystem. This subsystem is implemented using the Python library smtplib, which is configured for TLS-encrypted SMTP communication. The system captures the image of the unrecognized face, overlays a bounding box that highlights the recognized facial area, and saves the image in JPEG format. The technology appends a system timestamp, the recognition result, and the system status to an email. This email is sent simultaneously to at least two pre-registered addresses via the Raspberry Pi's integrated Wi-Fi module. This ensures that the homeowner is immediately notified of any break-ins, even in the event of a network outage.

[0048] In addition to intrusion detection, the system also includes fire and smoke monitoring. A semiconductor gas sensor MQ-2 is mounted in the sensor dock and electrically connected to the Raspberry Pi via an MCP3008 analog-to-digital converter, which operates via the SPI bus. The analog output of the MQ-2 varies depending on the presence of smoke or flammable gases. The Raspberry Pi continuously samples this output and converts it into concentration values ​​in ppm. The system compares these values ​​to a threshold of 300 ppm. Above this threshold, a transistor driver circuit simultaneously activates a piezoelectric buzzer and sends a remote IoT alert to the homeowner. The local alarm provides immediate notification to the occupants, while the IoT notification offers a remote alert if the house is unoccupied.

[0049] For energy management, the system integrates an ambient light sensor consisting of a light-dependent resistor in a voltage divider circuit. The divider output is digitized by the MCP3008 converter and read by the Raspberry Pi as lux equivalent values. The system uses control logic that switches the interior lighting on when the measured illuminance falls below 200 lux and off when it rises above 500 lux. To avoid unnecessary power consumption, this state is coupled with the output of a passive infrared sensor. The PIR sensor generates a high-level digital signal when it detects human movement. The system only activates the lighting relay when both the PIR sensor indicates occupancy and the light sensor indicates insufficient illumination. This dual state minimizes false triggers and ensures efficient energy use.

[0050] The Raspberry Pi is powered by a regulated 5-volt power supply, derived from a 12-volt DC power supply via a step-down converter. The converter incorporates an inductor, diode, and capacitor network to ensure stable regulation under variable loads. The Raspberry Pi also runs a watchdog timer daemon that continuously monitors system processes. If the system remains unresponsive for more than 60 seconds, the watchdog initiates an automatic reboot to restore normal operation.

[0051] Event management within the technology is supported by a logging subsystem. Every significant event—facial recognition, unknown person, fire hazard, light activation—is recorded in JSON format with associated timestamps, trust values, and system codes. These logs are encrypted with AES-128 before being stored locally. At defined intervals, the logs are transmitted via the IoT subsystem to a remote server for backup and auditability. This ensures that system activity can be reconstructed even in the event of local tampering.

[0052] The software architecture is implemented as a modular multiprocessing framework. Separate threads are created for camera image acquisition, hair cascade detection, facial recognition encoding, IoT alerting, fire and smoke monitoring, and ambient light control. A central scheduler on the Raspberry Pi manages these threads and prioritizes detection and safety-critical functions to ensure real-time response. This architecture allows the system to perform facial recognition and lock operation in parallel with environmental monitoring and IoT alerting without significant latency.

[0053] The combined structural hardware arrangement and technical implementation enable a comprehensive and autonomous home security system. By embedding facial recognition directly into the edge device, the system reduces reliance on cloud services, lowers latency, and protects user privacy. The integration of intrusion detection, fire monitoring, and energy-efficient lighting control into a single embedded device creates a robust, multifunctional solution for residential environments.

[0054] The system uses a Raspberry Pi 4B as its primary processor board. Integrated Wi-Fi, Ethernet, and sufficient processing speed with 4 GB of RAM were chosen for lightweight AI models. A USB camera connected via a communication port continuously records video images, which are processed using the OpenCV library. Haar Cascade classifiers embedded in the Python environment detect facial regions. Subsequently, the Dlib-based face recognition library encodes the detected faces into vectorized features. These features are then compared to the database stored locally on the Raspberry Pi.

[0055] For authorized detection, a signal is sent via GPIO control pins to activate the magnetic locking mechanism and unlock the door. In the event of unauthorized detection, the system simultaneously saves the captured image of the intruder locally and initiates SMTP-based communication to send a warning email to the homeowner's registered email address.

[0056] For safety integration, a fire / smoke detector module is connected to the Raspberry Pi's GPIO pins. If a detection threshold is exceeded, a dual response is triggered: a local buzzer sounds and an IoT alarm is sent. An infrared-based light and motion sensor subsystem provides additional automation, ensuring that interior lighting and fans are switched on and off depending on occupancy and ambient light levels.

[0057] The system is powered via a DC-DC step-down converter, which converts the 12V DC power supply in the home or vehicle to 5V, suitable for Raspberry Pi and sensors, thus ensuring continuous operation even with voltage fluctuations.

[0058] The entire system has a modular design, allowing individual subsystems to be easily replaced or upgraded. Performance has been validated under various environmental conditions, demonstrating fast response times, robustness in different lighting conditions, and minimal false positives.

[0059] The drawing and the preceding description show examples of embodiments. Those skilled in the art will recognize that one or more of the described elements can be combined to form a single functional element. Alternatively, certain elements can be divided into several functional elements. Elements of one embodiment can be added to another embodiment. For example, the sequence of the processes described here can be changed and is not limited to the manner described here. Furthermore, the actions of a flowchart need not be implemented in the sequence shown; nor does it necessarily have to be performed by all actions. Actions that are not dependent on other actions can also be performed in parallel with the other actions. The scope of the embodiments is in no way limited by these specific examples.Numerous variations are possible, whether explicitly stated in the specification or not, such as differences in structure, dimensions, and material use. The range of embodiments is at least as broad as specified in the following claims.

[0060] Advantages, further benefits, and problem solutions have been described above with regard to specific embodiments. However, the advantages, benefits, problem solutions, and all components that may lead to a particular advantage or solution occurring or becoming more apparent are not to be construed as critical, necessary, or essential features or components of any or all claims. REFERENCES 100 A quantum-based system for detecting fake news. 102 computing units 104 Webcam 106 Electromechanical door lock based on solenoids 108 A fire and smoke detection module 110 Ambient light sensor 112 Infrared occupancy sensor 114 Piezoelectric buzzer 116 Regulated Power Supply

Claims

[1] A system for home security and protection using artificial intelligence. The system includes: a computing unit configured with at least one quad-core processor and at least four gigabytes of RAM, the computing unit comprising an integrated wireless communication module and an Ethernet port; a webcam physically connected to a Universal Serial Bus interface of the computing unit, wherein the webcam is configured to continuously capture live image frames of a monitored access point; an electromagnetic electromechanical door lock that is mechanically mounted on a door structure and electrically coupled to a relay module, the relay module being in electrical connection with multi-purpose input / output pins of the computing unit; at least one fire and smoke detection module comprising a sensor coupled to an analog-to-digital converter, the output of which is electrically connected to the computing unit; an ambient light sensor comprising a light-dependent resistor arranged in a voltage divider circuit and coupled to the computing unit; an infrared occupancy detection sensor that is electrically connected to the computing unit; a piezoelectric buzzer that is electrically connected to the processing unit via a transistor driver circuit; and A regulated power supply with a DC-DC step-down converter configured to supply a direct current of five volts to the processing unit and associated sensors, the processing unit being configured to perform real-time facial recognition via the webcam input, actuate the magnetic door lock via the relay module upon detection of a known face, trigger the piezoelectric buzzer upon a hazard output from the fire and smoke sensor, and switch electrical loads connected via the relay module in response to outputs from the ambient light and infrared presence sensors. [2] System according to claim 1, wherein the magnetic lock comprises a cylindrical piston enclosed in a metal housing, wherein the piston is actuated into a retracted position when excited by a 5-volt DC signal provided via the relay module, wherein the relay module comprises a single-pole, single-stage switching element excited via the multi-purpose input / output pin of the computing unit, and wherein the magnetic lock is mounted such that, in the de-energized state, the piston engages with a strike plate attached to the door frame, thereby mechanically securing the door in a locked state. [3] System according to claim 1, wherein the fire and smoke detection module comprises a semiconductor sensor mounted on a printed circuit board, which contains a heating element and a tin dioxide-sensitive layer and is configured to change its resistance depending on the smoke or gas concentration, wherein the analog output of the sensor is passed through the analog-to-digital converter and transmitted as a digital voltage value to the computing unit, wherein the converter is electrically connected via a serial peripheral interface bus, thereby enabling continuous measurement of the ambient conditions. [4] System according to claim 1, wherein the ambient light sensor comprises a light-dependent resistor connected in series with a fixed-value resistor to form a voltage divider, wherein the divider is powered by the regulated 5-volt supply and wherein the output node of the divider is electrically connected to a channel of the analog-to-digital converter, wherein the digital output is processed by the computing unit to calculate corresponding illuminance levels, wherein the computing unit is also connected to interior lighting loads via a relay driver circuit, so that the loads are automatically switched on when the illuminance falls below a predetermined threshold. [5] System according to claim 1, wherein the infrared presence sensor comprises a passive infrared sensor module (PIR) housed in a plastic housing with a Fresnel lens, wherein the PIR sensor includes dual pyroelectric sensor elements configured to output a high-level digital signal when the infrared radiation patterns change due to human movement, wherein the output signal is passed directly to a digital input pin of the computing unit, and wherein the computing unit actuates electrical devices via the relay module only when both the output of the PIR sensor indicates occupancy and the output of the ambient light sensor indicates insufficient illumination. [6] System according to claim 1, wherein the regulated power supply comprises a step-down converter whose input terminal is connected to a 12-volt DC power source and whose output terminal is regulated to 5 volts DC. The step-down converter is mounted on a printed circuit board with an inductor, a Schottky diode, and an electrolytic capacitor to ensure voltage regulation. The regulated 5-volt output supplies power to the processing unit, the relay modules, the sensors, and the webcam, thus ensuring stable operation of the system under varying load conditions. [7] System according to claim 1, wherein the piezoelectric buzzer comprises a ceramic piezoelectric disk enclosed in a cylindrical housing, wherein the buzzer is connected in series with a transistor switch, wherein the base of the transistor is connected via a current limiting resistor to a general-purpose input / output pin of the computing unit, wherein the computing unit outputs a digital high signal to saturate the transistor and close the circuit. [8] System according to claim 1, wherein the relay module comprises an opto-isolated driver circuit with an optocoupler, a transistor and a freewheeling diode, wherein the module is configured to receive a low-level control signal from the computer and actuate an electromechanical relay coil, wherein the relay coil switches an isolated circuit connected to AC or DC household loads, thereby enabling safe electrical isolation between the control electronics and high-voltage equipment. [9] System according to claim 1, wherein the webcam comprises a CMOS image sensor module (Complementary Metal-Oxide-Semiconductor) with a lens with a fixed focal length, wherein the module is integrated into a housing that is mounted next to the door frame, wherein the USB output of the camera is directly connected to a USB input port of the computing unit. [10] System according to claim 1, wherein the computing unit is enclosed in a protective housing with ventilation slots and mounting devices, wherein the housing is mechanically attached to an inner wall next to the door, wherein the electrical wiring of magnetic lock, relay modules, sensors and power supply terminates in corresponding connections within the housing.

Images