SYSTEM AND METHOD FOR MONITORING USER USING PRIVACY-PRESERVING ACTIVITY RECOGNITION TO GENERATE ANOMALOUS BEHAVIOUR DETECTION

Abstract

ABSTRACT SYSTEM AND METHOD FOR MONITORING USER USING PRIVACY-PRESERVING ACTIVITY RECOGNITION TO GENERATE ANOMALOUS BEHAVIOUR DETECTION Disclosed is a monitoring system (100) for monitoring a user. The system (100) includes an input unit (102) and a processor (104). The input unit (102) receives first through third set of parameters of the user. The processor (104) recognizes, based on the first set of parameters, one or more activities associated with a schedule of the at least one user; determines, based on the second set of parameters, a deviation between the second set of parameters and a predefined second set of parameters; recognizes, based on the third set of parameters, an emotional wellbeing of the at least one user; and generates a notification signal based on the one or more activities, the deviation, and the emotional wellbeing. FIG. 1 is the reference figure

Specification

Description:FORM 2
THE PATENT ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003
COMPLETE SPECIFICATION
(See section 10; rule 13)

"SYSTEM AND METHOD FOR MONITORING USER USING PRIVACY-PRESERVING ACTIVITY RECOGNITION TO GENERATE ANOMALOUS BEHAVIOUR DETECTION"

Dr. Geetanjali V. Kale, an Indian citizen of, SCTR's Pune Institute of Computer Technology, Survey No. 27, Near, Trimurti Chowk, Bharati Vidyapeeth Campus, Dhankawadi, Pune, Maharashtra 411043
Mrs. Madhuri S. Wakode, an Indian citizen of, SCTR's Pune Institute of Computer Technology, Survey No. 27, Near, Trimurti Chowk, Bharati Vidyapeeth Campus, Dhankawadi, Pune, Maharashtra 411043
Mrs. Priyanka Niranjan Savadekar, an Indian citizen of, SCTR's Pune Institute of Computer Technology, Survey No. 27, Near, Trimurti Chowk, Bharati Vidyapeeth Campus, Dhankawadi, Pune, Maharashtra 411043
Mr. Niranjan Deokule, an Indian citizen of, SCTR's Pune Institute of Computer Technology, Survey No. 27, Near, Trimurti Chowk, Bharati Vidyapeeth Campus, Dhankawadi, Pune, Maharashtra 411043
Dr. Sanjay T. Gandhe, an Indian citizen of, SCTR's Pune Institute of Computer Technology, Survey No. 27, Near, Trimurti Chowk, Bharati Vidyapeeth Campus, Dhankawadi, Pune, Maharashtra 411043

THE FOLLOWING SPECIFICATION PARTICULARLY DESCRIBES THE INVENTION AND THE MANNER IN WHICH IT IS TO BE PERFORMED.
TECHNICAL FIELD

The present disclosure relates generally to monitoring. More specifically, the present disclosure relates to a system and a method for monitoring at least one user in a predefined space. More particularly, the present disclosure relates to a system and method for privacy-preserving human activity recognition and synopsis generation for anomalous behaviour detection.

BACKGROUND

Taking care of our loved ones who are aging demands constant monitoring of their daily activities and personal attention. Apart from monitoring physical health conditions, keeping them active, happy and emotionally stable is equally important and typically is more challenging. According to the Report of the Technical Group on Population Projections for India and States 2011-2036, there are nearly 138 million elderly persons in India in 2021 and is further expected to increase by around 56 million elderly persons in 2031. Person-centric digital healthcare solutions can help in healthy aging by increasing accessibility. Developing a smart and caring ambiance for old age people is one step ahead in this direction in developing countries like India. However there are various technical challenges in development and deployment of such ambience.

The existing IoT and surveillance based systems monitor physical health parameters and activities. However, many of these systems pay little attention to the emotional well-being of old people. Moreover, the privacy aspect is not considered in most of the systems. It is essential to keep the psychological conditions of elderly people regularized through their engagement in various physical and virtual activities. There is a requirement for developing a complete solution which monitors health conditions, recognizes major activities, reminds people about the major daily tasks, psychological well-being, attempts to improve it by suggesting activities and monitoring those. While doing so, the sensitive information related to a person shall be kept private. Additionally, storing video recording of activities demands high storage requirements. There is a need to develop short text synopsis of these videos to reduce the storage requirements and to preserve records for longer duration.

Further, conventional systems have insecurities related to privacy of data while uploading private data for analysis, which increases the chances of data breaches. No functionality to monitor emotional well-being along with health parameters. Lack of suggestion of activities to regularize the emotional state of elderly people. Lack of video synopsis in such applications to reduce the watching time and storage requirements. To solve the above problems there are few techniques that can be employed in the conventional systems. For example, homomorphic encryption, which is a cryptographic method that allows computations to be executed on encrypted data without needing decryption before computations. However, it is computationally intensive. Secure multi-party computation, which is a cryptographic protocol distributes computation across multiple parties, ensuring that no single party can access the data of the others. Differential privacy, which is a technique involves adding noise to user data to protect individual privacy during data analysis. It is computationally demanding and may be unsuitable for devices with limited memory.

Thus, there is a need for an advanced technical solution utilizing modern technological features that overcomes the aforementioned problems of conventional monitoring systems.

SUMMARY

In view of the foregoing, the present invention addresses multiple gaps in current IoT-based surveillance systems for elderly care by integrating emotional well-being monitoring, privacy preservation, activity recognition, and storage optimization.
Real-World Example: Consider a senior living community where each resident's health, daily activities, and emotional well-being are critical concerns. The staff often face challenges in manually monitoring everyone around the clock, and privacy becomes an issue with continuous video surveillance. Additionally, storing such vast amounts of video data strains the available memory capacity, making it difficult to retain information for long-term analysis.

The system includes the following:
Physical health monitoring: Using wearable devices like smart watches and sensors installed in the resident's environment, the system monitors heart rate, body temperature, mobility, and sleep patterns. Alerts are triggered for caregivers if any vital health parameter deviates from the norm.

Emotional well-being and activity suggestions: The system captures video and audio data at regular intervals, analyzes facial expressions and voice tones using local AI models to assess emotional well-being, and triggers activity recommendations like puzzles, music, or light exercises to improve mood. If a resident shows signs of distress, this is flagged for immediate caregiver intervention.

Privacy-preserving storage and analytics: Instead of storing continuous video, the system generates a short text synopsis summarizing major activities and emotional states, thus reducing storage needs and protecting privacy. Only the relevant segments of the video are stored for further analysis when anomalies or significant events are detected.

In an aspect, a privacy-preserving system for elderly health and emotion monitoring is provided. More specifically, a system for monitoring physical and emotional health parameters of elderly individuals is disclosed. The system includes a network of wearable sensors and ambient devices for collecting physical health data such as heart rate, temperature, and mobility, a multimodal emotion recognition module using cameras and microphones to analyze facial expressions, gestures, and voice for detecting the emotional well-being of the individual, a federated learning model deployed locally on user devices to process health and emotional data without transmitting it to a central server, ensuring privacy preservation, a suggestion module that provides activities to enhance psychological well-being based on the detected emotional state, and a video synopsis generation module that processes video and audio data, converting it into a summarized text or short video clip, thereby optimizing storage needs.

Real-World Example: In a smart home for elderly individuals, a resident named Mrs. Lee has wearable devices monitoring her heart rate, sleep patterns, and physical activity. At the same time, strategically placed cameras and microphones capture her facial expressions and voice at regular intervals. The system processes this data locally on her home's smart hub to assess her emotional and physical health without sharing sensitive information with external servers. One day, the system detects signs of emotional distress from her facial expressions and voice. The system suggests calming activities, like listening to soft music, and alerts her caregiver to provide additional support. All health and emotional data remain securely within Mrs. Lee's home environment, ensuring her privacy.

In another aspect, a method for privacy-preserving activity recognition and emotional regularization is disclosed. More specifically, a system for monitoring physical and emotional health parameters of elderly individuals is disclosed. The method includes collecting physical health parameters through wearable devices and ambient sensors, periodically capturing video and audio inputs to assess emotional states, processing the collected data locally using a federated learning approach, where machine learning models are trained on user data without sharing it with a central server, detecting significant anomalies in physical health and emotional well-being, followed by alerting caregivers, suggesting physical and virtual activities to improve emotional well-being based on recognized emotional states, generating a synopsis of the captured video for long-term storage and reference, preserving the privacy of the user.

Real-World Example: At a senior care center, Mr. Johnson's daily routine includes regular check-ups via a smart health monitoring system. The system continuously records his physical health parameters and captures short video segments to assess his emotional state. Recently, the system noticed deviations in his physical activity and emotional well-being. The system flagged these issues, sending an alert to caregivers, who found out that Mr. Johnson had been feeling isolated. The system then suggested social activities and games to engage him, and his emotional health improved within a week. Throughout the process, all data was analyzed on Mr. Johnson's device, protecting his personal information.

In another aspect, a system for optimizing storage requirements in a health and activity monitoring solution is disclosed. The system comprising a video-to-text conversion module that analyses video footage and generates a text-based synopsis summarizing the major activities and emotional states of elderly individuals, a speech-to-text conversion module that converts audio responses of elderly individuals to text for analysis and logging, a storage optimization feature that retains the most relevant segments of the captured video, discarding or summarizing the rest, thus reducing overall memory requirements.

Real-World Example: In a retirement community, continuous monitoring of elderly residents generates large amounts of video data. For example, Mrs. Gomez's activity is recorded 24/7, but most of the time, she is engaged in routine activities like reading or resting. Instead of storing entire days of video, the system generates a brief text-based synopsis summarizing significant events, such as her exercise routine, emotional responses, and any anomalies in her behavior. These text summaries, along with short video clips of important moments, are stored, significantly reducing the storage burden on the system while keeping records for future analysis.

In another aspect, a federated learning-based anomalous behavior detection system for detecting anomalous behavior in elderly individuals is disclosed. The system comprising a network of devices that capture audio, video, and health data, a local machine learning model trained using federated learning, where data is analyzed at the device level, maintaining privacy, an anomaly detection algorithm that flags deviations in normal health or behavior patterns, sending alerts to caregivers, and a collaborative training mechanism that allows models to improve continuously by learning from distributed datasets while ensuring that the raw data remains private.

Real-World Example: In an assisted living facility, each resident's room is equipped with cameras and sensors to monitor their behavior. The system detects anomalies, such as a resident showing signs of distress or not engaging in daily activities. The federated learning model processes this data locally in each resident's room, identifying patterns of concern without sending raw data to a central server. When the system detects that Mr. Davis has not moved from his chair for an unusually long period, it triggers an alert for caregivers. Since the learning model is distributed across multiple devices, it continually improves by learning from various residents' data without violating their privacy.

In another aspect, a video synopsis creation and activity recognition method generating synopses from recorded video data to reduce storage requirements in an elderly care monitoring system is disclosed. The method includes capturing continuous video streams of elderly individuals, identifying significant activities and behaviors through vision-based human activity recognition algorithms, summarizing the video by extracting key events and converting them into text or shorter video segments, reducing the total storage footprint, and preserving long-term records for expert analysis by storing only the summarized data, while discarding unnecessary portions of the video.

Real-World Example: In a memory care facility, Mrs. Patel's movements and activities are recorded by cameras to ensure her safety and monitor her cognitive health. The system employs advanced video processing to recognize important activities like walking, eating, or interacting with caregivers. It then creates a synopsis of these key moments, discarding unnecessary video footage. For example, the system summarizes Mrs. Patel's day with a short text synopsis indicating her meals, medication intake, and participation in a group activity, along with a brief video clip showing her engaged in conversation. This synopsis helps caregivers review her day efficiently, without sifting through hours of footage, and the summarized format reduces storage requirements.

The implementation of the system and method provided above has following technical effect and advantages:

Privacy Preservation: The use of federated learning ensures that personal data stays on the user's device, minimizing the risk of data breaches while still allowing models to improve over time through collaborative training.

Emotion Monitoring: By integrating emotional recognition into the system, it not only monitors physical health but also ensures the elderly individuals' emotional well-being, addressing a major gap in current surveillance systems.

Storage Efficiency: Video synopsis generation allows the system to store relevant and summarized data rather than entire video streams, making it easier to maintain long-term records without overwhelming storage resources.
Activity Recommendations: The system is proactive in suggesting activities to help regulate emotions, offering a companion-like experience for the elderly, which contributes to improved psychological health.

Scalability: The federated learning model allows for scaling the system across multiple users and geographical locations without compromising individual privacy.

BRIEF DESCRIPTION OF DRAWINGS

The above and still further features and advantages of aspects of the present disclosure becomes apparent upon consideration of the following detailed description of aspects thereof, especially when taken in conjunction with the accompanying drawings, and wherein:

FIG. 1A illustrates a block diagram of a system for monitoring at least one user in a predefined space, in accordance with an embodiment of the present disclosure;
FIG. 1B illustrates an architecture for the monitoring system, in accordance with an embodiment of the present disclosure;
FIG. 1C illustrates a room where the system is implemented, in accordance with an exemplary embodiment of the present disclosure;
FIG. 2 illustrates an architecture for the proposed Federated Learning technique, in accordance with an embodiment of the present disclosure; and
FIG. 3 illustrates a flow chart of a method for monitoring at least one user in a predefined space, in accordance with an embodiment of the present disclosure.
To facilitate understanding, like reference numerals have been used, where possible, to designate like elements common to the figures.






DETAILED DESCRIPTION

The subject matter of example aspects, as disclosed herein, is described with specificity to meet statutory requirements. However, the description itself is not intended to limit the scope of this disclosure. Rather, the inventor/inventors have contemplated that the claimed subject matter might also be embodied in other ways, to include different features or combinations of features similar to the ones described in this document, in conjunction with other technologies. Generally, the various aspects including the example aspects relate to a system and a method for monitoring at least one user in a predefined space.

The system of the present disclosure employs Federated Learning (FL) technique that is best suited for our problem statement to analyse the data on site without sending it to a central server. In federated learning, an aggregation algorithm averages the parameters obtained from local models present on end user's devices and these weights are then used for training the global model present on the central server. Users' data privacy is ensured as data will not leave the client nodes for training purposes.

FIG. 1A illustrates a block diagram of a monitoring system 100 for monitoring at least one user in a predefined space, in accordance with an embodiment of the present disclosure. The system 100 may be configured to monitor at least one user present in a predefined space. The system 100 focuses on design and implementation of novel algorithms to recognize human activities, determine psychological conditions and generate video synopsis with privacy preservation using federated learning.

The monitoring system 100 may include an input unit 102, a processor 104, a database 108, and an output unit 110. The input unit 102, the processor 104, the database 108, and the output unit 110 may be coupled to each other. Specifically, the input unit 102, the processor 104, the database 108, and the output unit 110 may be communicatively coupled to each other by way of a communication channel 112. The communication channel 112 may be configured to facilitate exchange of information among various components associated with the system 100. Specifically, the communication channel 112 may be configured to facilitate exchange of information among the input unit 102, the processor 104, the database 108, and the output unit 110.

The input unit 102 may be configured to receive a first set of parameters, a second set of parameters, and a third set of parameters of the at least one user. The processor 104 coupled to the input unit 102. The processor 104 may be configured to recognize, based on the first set of parameters, one or more activities associated with a schedule of the at least one user. The processor 104 may be configured to recognize the one or more activities by way of one or more Federated Learning (FL) techniques. The processor 104 may be further configured to determine, based on the second set of parameters, a deviation between the second set of parameters and a predefined second set of parameters. The processor 104 may be configured to determine the deviation by way of the one or more FL techniques. The processor 104 may be further configured to recognize, based on the third set of parameters, an emotional wellbeing of the at least one user. The processor 104 may be configured to recognize the emotional wellbeing by way of the one or more FL techniques. The processor 104 may be configured to generate a notification signal based on the one or more activities, the deviation, and the emotional wellbeing.

In some embodiments of the present disclosure, the input unit 102 may include a user interface 106a, a plurality of sensors 106b, a plurality of cameras 106c, and microphones 106d. The user interface 106a may be configured to receive the first set of parameters. The first set of parameters comprising daily routine events of the at least one user. The plurality of sensors 106b may be configured to capture the second set of parameters. The second set of parameters comprising physiological parameters of the at least one user. The plurality of cameras 106cand the microphones 106d may be configured to capture the third set of parameters. The third set of parameters comprising audio parameters and video parameters of the at least one user. Specifically, the plurality of cameras 106c may be configured to capture the video parameters. The microphones 106d may be configured to capture the audio parameters.

In some embodiments of the present disclosure, the processor 104 is further configured to generate a plurality of video synopsis that correspond to video parameters that are captured over a predefined interval of time.

In some embodiments of the present disclosure, the monitoring system 100 further comprising the database 108. The database 108 may be coupled to the processor 104. The database 108 may be configured to store the first set of parameters, the second set of parameters, the audio parameters, and the plurality of video synopsis. Specifically, the database 108 may include a plurality of repositories such that the first set of parameters, the second set of parameters, the audio parameters, and the plurality of video synopsis are stored in respective repositories.

In some embodiments of the present disclosure, the monitoring system 100 further comprising an output unit 110 coupled to the input unit 102 and the processor 104, wherein the notification signal is executed by the output unit 110.

In some embodiments of the present disclosure, the output unit 110 may be configured to display the one or more activities. The output unit 110 may be further configured to generate one or more voice commands in response to the one or more activities.

In some embodiments of the present disclosure, the output unit 110 may be configured to transmit one or more alerts to concerned authorities. Specifically, the output unit 110 may be configured to transmit the oneor more alerts to concerned authorities when the deviation is beyond a predefined threshold. In some preferred examples of the present disclosure, the concerned authorities may include healthcare professionals, one or more persons related to the at least one user, and the like. Embodiments of the present disclosure are intended to include and/or otherwise cover any type of the concerned authorities, without deviating from the scope of the present disclosure.

In some embodiments of the present disclosure, the output unit 110 may be configured to, when the emotional wellbeing is unfavourable, trigger predetermined activities to regularise the emotional wellbeing.

Working example: The system being implemented in an elderly home monitoring system. In an elderly care home, a monitoring system (100) is deployed to assist with tracking the daily activities, physiological health, and emotional well-being of the residents. The system uses various sensors, cameras, and microphones installed in each resident's room to gather data while ensuring privacy through federated learning techniques.

Input Unit (102): The input unit (102) collects three different sets of parameters for each resident in the elderly care home:

First Set of Parameters (Daily Routine Events):Residents provide their daily schedule through a user interface (106a) such as a smartphone app or a touchscreen display in their room. For example, Mrs. Johnson inputs her regular daily routine: breakfast at 8 AM, a walk in the park at 10 AM, lunch at 12 PM, and so on.

Second Set of Parameters (Physiological Parameters): Wearable sensors (106b), such as smartwatches and body temperature sensors, continuously monitor physiological parameters like heart rate, body temperature, and sleep quality. For example, Mr. Lee wears a smartwatch that tracks his pulse rate and oxygen levels.

Third Set of Parameters (Audio and Video for Emotion Monitoring):Cameras (106c) and microphones (106d) are placed in the resident's room to capture audio and video data at scheduled intervals. This data helps recognize facial expressions and vocal tone to assess emotional well-being. For example, the camera detects if Mrs. Gomez is smiling or showing signs of stress, while the microphone captures her voice as she talks to her family over the phone.

Processor (104) with Federated Learning (FL): The processor (104) plays a central role in interpreting the data collected from various sensors, cameras, and microphones. Here's how it works in real-time:

Activity Recognition Based on First Set of Parameters: Using federated learning techniques, the processor (104) compares Mrs. Johnson's actual activities with her pre-defined schedule. For instance, if she hasn't gone for her usual walk at 10 AM, the system recognizes this as an anomaly and generates an alert for the caregiver.

Deviation Detection Based on Second Set of Parameters: The processor compares real-time physiological data with predefined normal ranges. For instance, if Mr. Lee's pulse rate is unusually high compared to his normal rate, the system flags this as a deviation, alerts medical staff, and requests immediate attention.

Emotion Recognition Based on Third Set of Parameters: By processing audio and video data through federated learning models, the system recognizes changes in emotional well-being. For example, if Mrs. Gomez's facial expression shows distress or sadness over a period, the system detects it and triggers pre-set activities (like music therapy or a phone call to family) to improve her emotional state.
Video Synopsis Creation (104): The system generates video synopses for each resident, summarizing key activities throughout the day, rather than storing full-length videos. For example, rather than storing hours of footage of Mrs. Johnson's activities, the system creates a short synopsis showing her morning walk, meal times, and social interactions with other residents. This reduces storage needs while still providing valuable insights into her daily routine.

Database (108): The database (108) stores all collected data including:
Mrs. Johnson's daily routine (first set of parameters).
Mr. Lee's physiological health records from wearable devices (second set of parameters).

Video summaries and emotional health records based on audio/video analysis of residents like Mrs. Gomez (third set of parameters).

This data is accessible to caregivers and health professionals for review and long-term care planning.

Output Unit (110): The output unit (110) is responsible for sending notifications and alerts based on the data processed by the system:
Activity-Based Notifications and Commands: If Mrs. Johnson misses her scheduled walk, the system sends a notification to her smartphone or room display. Additionally, it can generate voice commands reminding her of the activity, such as: "It's time for your walk in the park, Mrs. Johnson."

Alerts to Caregivers for Deviation: If Mr. Lee's physiological parameters deviate from the normal range (e.g., a sudden drop in oxygen levels), the system sends an urgent alert to the concerned medical team or caregivers via the facility's monitoring dashboard or mobile app.
Triggered Activities for Emotional Well-Being: If Mrs. Gomez is feeling emotionally low, the system automatically triggers activities such as playing her favorite music or suggesting she call her family. The output unit could also display motivational messages on the screen in her room or initiate a video call.

Federated Learning and Privacy: A major advantage of this system is that all data analysis occurs locally on the residents' devices through federated learning (FL). For example, the emotional well-being recognition for Mrs. Gomez is done on the device installed in her room, ensuring her sensitive data never leaves the facility's secure environment. Caregivers receive alerts and summaries, but no raw data is sent to a central server, significantly enhancing privacy.

Summary of the Working System Flow:
Data Collection: Residents like Mrs. Johnson, Mr. Lee, and Mrs. Gomez provide input (daily routines, physiological health, and emotional data) through wearable sensors, cameras, and a user interface.

Data Processing with FL:The processor (104) recognizes activities, detects deviations in health, and analyzes emotional well-being using federated learning techniques, all while maintaining privacy.

Notification Generation: Based on recognized deviations or emotional states, the system generates notifications for the residents (reminders or motivational messages) and alerts for caregivers if needed.

Video Synopsis: Summarized videos and data logs are created for easy review, reducing storage demands while keeping a concise record of each resident's daily activities.
Action Execution: If a deviation or emotional issue is detected, appropriate actions are triggered-whether it's sending alerts to medical personnel or suggesting activities to improve emotional well-being.

The system ensures constant monitoring of physical health, daily activities, and emotional well-being, while safeguarding privacy through federated learning and optimizing storage through video synopsis creation.

FIG. 1B illustrates an architecture 114 for the monitoring system 100, in accordance with an embodiment of the present disclosure. The plurality of sensors, the microphones, the wearable gadgets (devices), and the cameras may be configured to capture the one or more parameters of the at least one user. Specifically, the plurality of sensors may be configured to capture the one or more physiological parameters of the at least one user. The cameras may be configured to make a video of the at least one user for the predefined interval of time. Specifically, the camera may facilitate to capture the one or more activities that may be performed by the at least one user in the predefined space. The microphones may be configured to capture the audio parameters of the at least one user. Specifically, the microphones may be configured to record one or more responses that may be given by the at least one user corresponding to one or more questionnaires.

The input unit 102 may be configured to receive the one or more parameters that may be captured or recorded by the plurality of sensors, the microphones, the wearable gadgets, and the cameras. Specifically, the input unit 102 may be configured to channelize the one or more parameters for further processing of the one or more parameters by the processor 104.

The processor 104 may be configured to receive the one or more parameters of the at least one user from the input unit 102. The processor 104 may be further configured to process the one or more parameters that may be received from the user. The processor 104 may be configured to facilitate significant activity monitoring associated with the at least one user. Specifically, the processor 104 may be configured to facilitate significant activity monitoring based on the one or more parameters. The processor 104 may be configured to monitor the health of the at least one user based on the one or more parameters. The processor 104 may be further configured to recognize the one or more emotions or the emotional wellbeing of the at least one user. Specifically, the processor 104 may be configured to recognize the one or more emotions or the emotional wellbeing based on the one or more parameters. The processor 104 may be further configured to determine psychological wellbeing of the at least one user. Specifically, the processor 104 may be configured to determine the psychological wellbeing of the at least one user based on the one or more parameters. The processor 104 may be further configured to generate the notification signal based on the one or more parameters.

The output unit 110 may be configured to execute the notification signal that may be generated by the processor 104. The output unit 110 may be configured to facilitate to provide timely notification to specify an activity of the one or more activities associated with the at least one user. Specifically, the output unit 110 may be configured to provide timely notification to specify the activity on a smart phone and voice reminders. The output unit 110 may be further configured to transmit an alert or a notification to concerned authorities. Specifically, the output unit 110 may be configured to transmit the alert or the notification to the concerned authorities, when the processor 104 determines an anomaly in the one or more parameters. The processor 104 may be configured to determine the anomaly in the one or more parameters by comparing the one or more parameters with a set of predefined parameters. The processor 104, in order to determine the anomaly, may be configured to determine the deviation between the one or more parameters and the set of predefined parameters. The processor 104 may be configured to determine the anomaly when the deviation lies beyond the threshold.
The output unit 110 may be further configured to create or generate the plurality of synopsis. Thus, the generation of the plurality of synopsis reduces or eliminates the troubles associated to storage of long videos. The plurality of synopsis may be stored in the database 108. Therefore, there is no requirement of storing the long videos in the database 108 and thereby solving the issues associated with the storage space of the database 108. The output unit 110 may be further configured to generate logs. Specifically, the output unit 110 may be configured to generate the logs based on the one or more parameters.

FIG. 1C illustrates a room 116 where the system 100 is implemented, in accordance with an exemplary embodiment of the present disclosure. The room 116 may include one or more facilities that may be used by the at least one user. Specifically, the one or more facilities may facilitate the at least one user to perform the one or more activities. The user interface 106a may be placed or disposed in the room 116. Specifically, the user interface 106a may be placed or disposed in the room 116 at a location where the at least one user may easily interact with the system 100. In some examples of the present disclosure, the user interface 106a may be a part of a user device that may be held by the at least one user. In some other examples of the present disclosure, the user device may be held by the at least one user such that the at least one user provides the one or more parameters to the input unit 102 through the user interface 106a. The plurality of sensors 106b, the plurality of cameras 106c, and the microphones 106d may be distributed at different locations of the room 116. Specifically, the plurality of sensors 106b, the plurality of cameras 106c, and the microphones 106d that may be distributed at the different locations of the room 116 may be configured to capture or record the one or more parameters of the at least one user. The plurality of sensors 106b, the plurality of cameras 106c, and the microphones may be preferably installed or deployed at the locations of the room where the at least one user stays for longer period of time for a particular day. The input unit 102 may be configured to receive the captured or recorded one or more parameters from the plurality of sensors 106b, the plurality of cameras 106c, and the microphones. The input unit 102 may be configured to receive a first set of parameters, a second set of parameters, and a third set of parameters of the at least one user. The processor 104 coupled to the input unit 102. The processor 104 may be configured to recognize, based on the first set of parameters, one or more activities associated with a schedule of the at least one user. The processor 104 may be configured to recognize the one or more activities by way of one or more Federated Learning (FL) techniques. The processor 104 may be further configured to determine, based on the second set of parameters, a deviation between the second set of parameters and a predefined second set of parameters. The processor 104 may be configured to determine the deviation by way of the one or more FL techniques. The processor 104 may be further configured to recognize, based on the third set of parameters, an emotional wellbeing of the at least one user. The processor 104 may be configured to recognize the emotional wellbeing by way of the one or more FL techniques. The processor 104 may be configured to generate a notification signal based on the one or more activities, the deviation, and the emotional wellbeing. The output unit 110 may be configured to display the one or more activities. The output unit 110 may be further configured to generate one or more voice commands in response to the one or more activities. The output unit 110 may be configured to transmit one or more alerts to concerned authorities. Specifically, the output unit 110 may be configured to transmit the one or more alerts to concerned authorities when the deviation is beyond a predefined threshold. The output unit 110 may be configured to, when the emotional wellbeing is unfavourable, trigger predetermined activities to regularise the emotional wellbeing.

FIG. 2 illustrates an architecture 200 for the proposed Federated Learning technique, in accordance with an embodiment of the present disclosure. Specifically, FIG. 2 illustrates the architecture 200 for the proposed FL technique that is implemented in the monitoring system 100 of FIG. 1A. The FL technique is based on empowered privacy preserving system for human activity and emotion recognition with synopsis generation. The Federated Learning (FL) technique represents an emerging research domain within privacy-preserving machine learning (ML). In traditional ML paradigms, model training occurs on a central server where data from all participants is aggregated. Conversely, FL involves local analysis of data on individual user devices rather than transmitting it to a centralized server, such as cloud storage. This ensures that user data remains private and confined to the nodes or devices where it originates. In the proposed system, a model-to-data approach will be adopted, wherein recognition and detection algorithms will be trained locally, proximate to the data, rather than on centralized servers.

FIG. 3 illustrates a flow chart of a method 300 for monitoring at least one user in a predefined space, in accordance with an embodiment of the present disclosure. The method 300 may further include following steps for monitoring the at least one user present in the predefined space.

At step 302, the system 100 may be configured to receive the first set of parameters, a second set of parameters, and a third set of parameters of the at least one user. Specifically, the system 100, by way of the input unit 102, may be configured to receive the first set of parameters, a second set of parameters, and a third set of parameters of the at least one user.

At step 304, the system 100 may be configured to recognize the one or more activities associated with a schedule of the at least one user. Specifically, the system 100, by way of the processor 104 coupled to the input unit 102, may be configured to recognize the one or more activities associated with the schedule of the at least one user. The one or more activities are recognized based on the first set of parameters. The processor 104 is configured to recognize the one or more activities by way of one or more Federated Learning (FL) techniques.
At step 306, the system 100 may be configured to determine the deviation between the second set of parameters and a predefined second set of parameters. Specifically, the system 100, by way of the processor 104, may be configured to determine the deviation between the second set of parameters and the predefined second set of parameters. The deviation is determined based on the second set of parameters, wherein the processor 104 is configured to determine the deviation by way of the one or more FL techniques.

At step 308, the system 100 may be configured to recognize the emotional wellbeing of the at least one user. Specifically, the system 100 may be configured to, by way of the processor 104, recognize the emotional wellbeing of the at least one user. The emotional wellbeing is recognized based on the third set of parameters, wherein the processor 104 is configured to recognize the emotional wellbeing by way of the one or more FL techniques.

At step 310, the system 100 may be configured to generate a notification signal. Specifically, the system 100, by way of the processor 104, may be configured to generate the notification signal based on the one or more activities, the deviation, and the emotional wellbeing.

At step 312, the system 100 may be configured to execute the notification signal. Specifically, the system 100, by way of the output unit 110 coupled to the input unit 102 and the processor 104, may be configured to execute the notification signal.

Thus, the monitoring system 100 of the present disclosure is advantageously a secure system that is based on a methodology employing innovative privacy-preserving techniques to safeguard user's confidential physical and health data during analysis through distributed learning. The system 100 is implemented as a multi-modal system for monitoring and recognizing user activities by analysing physical and emotional data. The system 100 facilitates a privacy-preserving companion system that delivers personalized recommendations to enhance users' emotional well-being by improving their emotional states. The system 100 provides an efficient and secure storage mechanism for video data collected from surveillance cameras, utilizing video synopsis and summarization techniques.

The foregoing discussion of the present disclosure has been presented for purposes of illustration and description. It is not intended to limit the present disclosure to the form or forms disclosed herein. In the foregoing Detailed Description, for example, various features of the present disclosure are grouped together in one or more aspects, configurations, or aspects for the purpose of streamlining the disclosure. The features of the aspects, configurations, or aspects may be combined in alternate aspects, configurations, or aspects other than those discussed above. This method of disclosure is not to be interpreted as reflecting an intention the present disclosure requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed aspect, configuration, or aspect. Thus, the following claims are hereby incorporated into this Detailed Description, with each claim standing on its own as a separate aspect of the present disclosure.

Moreover, though the description of the present disclosure has included description of one or more aspects, configurations, or aspects and certain variations and modifications, other variations, combinations, and modifications are within the scope of the present disclosure, e.g., as may be within the skill and knowledge of those in the art, after understanding the present disclosure. It is intended to obtain rights which include alternative aspects, configurations, or aspects to the extent permitted, including alternate, interchangeable and/or equivalent structures, functions, ranges or steps to those claimed, whether or not such alternate, interchangeable and/or equivalent structures, functions, ranges or steps are disclosed herein, and without intending to publicly dedicate any patentable subject matter.  , Claims:We Claim:

1. A monitoring system (100) for monitoring at least one user present in a predefined space, the monitoring system (100) comprising:
an input unit (102) configured to receive a first set of parameters, a second set of parameters, and a third set of parameters of the at least one user;

a processor (104) coupled to the input unit (102), and configured to:
recognize, based on the first set of parameters, one or more activities associated with a schedule of the at least one user, wherein the processor (104) is configured to recognize the one or more activities by way of one or more Federated Learning (FL) techniques;

determine, based on the second set of parameters, a deviation between the second set of parameters and a predefined second set of parameters, wherein the processor (104) is configured to determine the deviation by way of the one or more FL techniques;

recognize, based on the third set of parameters, an emotional wellbeing of the at least one user, wherein the processor (104) is configured to recognize the emotional wellbeing by way of the one or more FL techniques; and

generate a notification signal based on the one or more activities, the deviation, and the emotional wellbeing.






2. The monitoring system (100) as claimed in claim 1, wherein the input unit (102) comprising:
a user interface (106a) configured to receive the first set of parameters, wherein the first set of parameters comprising daily routine events of the at least one user;
a plurality of sensors (106b) configured to capture the second set of parameters, wherein the second set of parameters comprising physiological parameters of the at least one user; and
a plurality of cameras (106c) and microphones (106d) configured to capture the third set of parameters, wherein the third set of parameters comprising audio parameters and video parameters of the at least one user.

3. The monitoring system (100) as claimed in claim 1, wherein the processor (104) is further configured to generate a plurality of video synopsis that correspond to video parameters that are captured over a predefined interval of time.

4. The monitoring system (100) as claimed in claim 2, wherein the system further comprising a database (108) to store the first set of parameters, the second set of parameters, the audio parameters, and the plurality of video synopsis.

5. The monitoring system (100) as claimed in claim 1, wherein the system further comprising an output unit (110) coupled to the input unit (102) and the processor (104), wherein the notification signal is executed by the output unit (110).

6. The monitoring system (100) as claimed in claim 5, wherein the output unit (110) is configured to display the one or more activities and generate one or more voice commands in response to the one or more activities.

7. The monitoring system (100) as claimed in claim 5, wherein the output unit (110) is configured to transmit one or more alerts to concerned authorities when the deviation is beyond a predefined threshold.

8. The monitoring system (100) as claimed in claim 5, wherein the output unit (110) is configured to, when the emotional wellbeing is unfavourable, trigger predetermined activities to regularise the emotional wellbeing.

9. A method (300) for monitoring at least one user present in a predefined space, the method (300) comprising:

receiving (302), by way of an input unit (102), a first set of parameters, a second set of parameters, and a third set of parameters of the at least one user;

recognizing (304), by way of a processor (104) coupled to the input unit (102), one or more activities associated with a schedule of the at least one user, wherein the one or more activities are recognized based on the first set of parameters, wherein the processor (104) is configured to recognize the one or more activities by way of one or more Federated Learning (FL) techniques;

determining (306), by way of the processor (104), a deviation between the second set of parameters and a predefined second set of parameters, wherein the deviation is determined based on the second set of parameters, wherein the processor (104) is configured to determine the deviation by way of the one or more FL techniques;

recognizing (308), by way of the processor (104), an emotional wellbeing of the at least one user, wherein the emotional wellbeing is recognized based on the third set of parameters, wherein the processor (104) is configured to recognize the emotional wellbeing by way of the one or more FL techniques; and
generating (310), by way of the processor (104), a notification signal based on the one or more activities, the deviation, and the emotional wellbeing.

10. The method (300) as claimed in claim 9, further comprising executing (312), by way of an output unit (110) coupled to the input unit (102) and the processor (104), the notification signal.

Documents