System and method for real-time exercise tutoring

The DFTA system addresses the lack of personalized, real-time exercise form analysis by using on-device computer vision and voice feedback, ensuring accurate exercise guidance and reducing injury risk on resource-constrained devices.

US20260355574A1Pending Publication Date: 2026-11-12AAWAAZ INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
US19/443001
Authority / Receiving Office
US · United States
Patent Type
Applications(United States)
Current Assignee / Owner
Priority Date
2025-05-12
Filing Date
2026-01-07
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Existing fitness solutions fail to provide accurate, real-time exercise form analysis tailored to individual needs, skill levels, and specific exercise requirements, often requiring gym equipment or external sensors, and do not address exercise form issues due to physical limitations or injuries, limiting inclusivity and increasing injury risk.

Method used

A dynamic fitness training assistant (DFTA) implemented on user devices, utilizing computer vision and voice-guided feedback, performs real-time pose estimation and anomaly detection, providing personalized exercise plans and feedback without the need for specialized equipment, and adapting to user-specific physical constraints.

Benefits of technology

Enables accurate, real-time exercise form analysis and feedback, enhancing user inclusivity and reducing injury risk by integrating on-device systems with voice and visual guidance, suitable for resource-constrained devices like smartphones.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure US20260355574A1-D00000_ABST
    Figure US20260355574A1-D00000_ABST
Patent Text Reader

Abstract

A method for real-time exercise tutoring is disclosed. The method employs a dynamic fitness training assistant (DFTA) implemented on a user device. The DFTA is prepared for the real-time exercise tutoring by setting up all necessary components. The DFTA captures a live video of a user using a front camera of the user device, analyses the captured live video to detect key points of the user. The DFTA verifies full-body visibility of the user and overlays the detected key points onto the live video of the user. The DFTA creates a dynamic virtual workout box around the live video of the user in real-time based on the detected key points. The DFTA pauses and resumes data analysis based on determining whether the user is within the dynamic virtual workout box, and monitors and analyses the user movement using the detected key points and calculated joint angles.
Need to check novelty before this filing date? Find Prior Art

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of the provisional patent application titled “System and Method for Real-Time Exercise Tutoring”, application number 202541045584, filed in the Indian Patent Office on May 12, 2025. The specification of the above-referenced patent application is incorporated herein by reference in its entirety.BACKGROUND

[0002] Proper exercise technique is critical for preventing injuries and achieving optimal fitness results. Existing fitness solutions often fail to deliver accurate, real-time exercise form analysis tailored to individual needs, skill levels, and specific exercise requirements. Additionally, many solutions depend on gym equipment, external weights, and external sensors like smart watches, making them less accessible for everyday users. Hence, there is long-felt need for a method and a system that delivers accurate, real-time exercise form / posture analysis tailored to individual needs, skill levels, and specific exercise requirements, especially on resource-constrained devices such as mobiles phones. Furthermore, there is long-felt need for a method and a system that provides accurate, real-time exercise form analysis of user's exercise form during bodyweight workouts, eliminating the need for gym access or external equipment, while enabling users to perform effective, personalized workouts anytime, anywhere.

[0003] Traditional pose estimation tools require significant computational resources or rely on human oversight, making them impractical for widespread use. Furthermore, these systems rarely address scenarios where a user's incorrect exercise form / posture is not due to a lack of knowledge or improper technique but rather stems from physical limitations, injuries, or anomalies in body structure of the user. This oversight limits inclusivity and increases the risk of injury. Hence, there is long-felt need for a method and a system that integrates a fully automated, on-device system with voice-guided and visual feedback, and anomaly detection, for providing a robust, inclusive, and real-time exercise mentor for users worldwide without the need for specialized equipment.BRIEF DESCRIPTION OF THE DRAWINGS

[0004] The foregoing summary, as well as the following detailed description of the invention, is better understood when read in conjunction with the appended drawings. For illustrating the embodiments herein, exemplary constructions of the embodiments are shown in the drawings. However, the embodiments herein are not limited to the specific methods and components disclosed herein. The description of a method step or a component referenced by a numeral in a drawing is applicable to the description of that method step or that component shown by that same numeral in any subsequent drawing herein.

[0005] FIGS. 1A-1B illustrate a method for real-time exercise tutoring.

[0006] FIGS. 2A-2L exemplarily illustrate screen shots of an embodiment of a dynamic fitness training assistant (DFTA) displayed on a user device during real-time exercise tutoring.

[0007] FIG. 3 illustrates an example of a dynamic fitness training assistant (DFTA) implemented on a user device.DETAILED DESCRIPTION OF THE INVENTION

[0008] Various aspects of the disclosure herein are embodied as a system, a method, or a non-transitory, computer-readable storage medium having one or more computer readable program codes stored thereon. Accordingly, various embodiments of the disclosure herein take the form of an entirely hardware embodiment, an entirely software embodiment comprising, for example, microcode, firmware, software, etc., or an embodiment combining software and hardware aspects that are referred to herein as a “system”, a “module”, an “engine”, a “circuit”, or a “unit”. As used herein, the terms “exercise form” or “exercise posture” refer to the spatial configuration, position, and orientation of a user's body parts relative to one another and / or to an external reference point or apparatus during the performance of a physical movement or exercise. For the purposes of this invention, the terms “exercise form”, “exercise posture”, “form”, and “posture” are used interchangeably to refer to the dynamic or static spatial arrangement of a user's anatomical segments (including limbs, torso, and head) relative to each other or to an external reference during the performance of a physical activity.

[0009] FIGS. 1A-1B illustrate a method for real-time exercise tutoring. The method 100 employs a dynamic fitness training assistant (DFTA) 303, shown in FIG. 3, executable by at least one processor 301 of a user device 200, shown in FIG. 2. The at least one processor 301 is configured to execute computer program instructions stored in a memory 302 of the user device 200 for performing the method 100. In an embodiment, the DFTA 303 is implemented as an on-device software installed and run on a user device 200. The user device 200 is an electronic device, for example, one or more of a personal computer, a tablet computing device, a mobile computer, a mobile phone, a smartphone, a portable computing device, a laptop, a personal digital assistant, a wearable computing device such as smart glasses, a smart watch, etc., a touch centric device, a workstation, a client device, a portable electronic device, a network-enabled computing device, an interactive network-enabled communication device, an image capture device, any other suitable computing equipment, combinations of multiple pieces of computing equipment, etc. In another embodiment, the DFTA 303 is implemented as a cloud-based platform hosted on a server or a network of servers accessible via a network. In another embodiment, the DFTA 303 is implemented in a cloud computing environment. As used herein, “cloud computing environment” refers to a processing environment comprising configurable computing, physical, and logical resources, for example, networks, servers, storage media, virtual machines, applications, services, etc., and data distributed over a network, for example, the internet. The cloud computing environment is one of a private cloud, a public cloud, a hybrid cloud, etc. The cloud computing environment provides an on-demand network access to a shared pool of the configurable computing physical and logical resources.

[0010] Upon initiation of the dynamic fitness training assistant (DFTA) 303 on the user device 200, a resource constrained device, such as, a smartphone, by clicking the DFTA icon on a graphical user interface (GUI) 201 of the user device 200. Upon initial use, the DFTA 303 prompts the user to input personal information, such as, age, weight, height, fitness level, etc., for user-specific configuration. The DFTA 303 utilizes the user's personal information to generate personalized and curated exercise plans tailored to each individual's needs. Before beginning a workout, the DFTA 303 presents a user 304 with a curated list of exercises tailored to the user's daily fitness requirements. For each exercise, the user 304 can watch an animated video that explains how to perform the exercise, including the type of exercise such as, warm up, full body, upper body, or lower body, a description of the exercise, set and repetition count, estimated caloric expenditure based on the user's requirements, and a “Start” button to begin the workout session. The user 304 positions the user device 200 on a stable surface, such as, a floor, with the user device 200 positioned approximately perpendicular to the ground to achieve an optimal field of view for motion tracking. Once the user device 200 is properly positioned, the user 304 starts the workout session by clicking start button provided by the DFTA 303.

[0011] Upon activation, the DFTA 303 automatically configures the necessary components on the user device 200 to facilitate real-time exercise tutoring. These necessary components include initializing variables, setting up the graphical user interface (GUI) 201, configuring the front camera, verifying permissions, enabling voice feedback, connecting to a cloud-based server for user-specific data, and other hardware or software components required for real-time motion capture and analysis. The user-specific data comprises personal information of the user 304 comprising age, weight, height, injuries, health, and fitness level of the user 304. If any of the necessary settings or permissions are not enabled, the DFTA 303 generates a prompt requesting the user 304 to grant the necessary permissions. For instance, initializing variables on the user device 200 sets up all internal variables needed for tracking exercises and monitoring user actions and providing visual and audio feedback. For example, initializing counters for repetitions, states, for example, rest vs active, and angle thresholds. For instance, the graphical user interface (GUI) 201 is initialized to prepare and display relevant elements, such as, “Start Exercise” button, exercise instructions, real-time feedback, feedback messages, rep counts, and a summary panel for session data, on the user device 200. The GUI 201 is further configured to show a live camera feed with key points or landmarks overlayed on a live video 202 of the user 304. For instance, voice feedback is configured as a controlled output such that only one audio instruction is active at a time, and subsequent feedback is deferred until completion of the ongoing audio instruction.

[0012] The dynamic fitness training assistant (DFTA 303) requires access to the user device's camera to capture real-time frames for pose detection and access to the user device's storage to save session data. If the necessary permissions are not granted, the DFTA 303 will generate a prompt requesting the user 304 to grant the necessary permissions before proceeding further. The camera permission is used to activate the front-facing camera, providing a live feed for exercise tracking. The storage permission allows the DFTA 303 to save workout session data, including metrics such as, correct / incorrect reps, performance summaries, etc., for future analysis to give curated exercises based on performance and find anomalies in a body structure of the user 304. If permissions are denied, the DFTA 303 prompts the user 304 with a message for example, “Camera and storage permissions are required to track your exercise. Please enable them.”, “Granting storage permission is not required for real-time posture analysis or feedback generation, which are executed locally on the user device.”.

[0013] Before initiating an exercise session, the dynamic fitness training assistant (DFTA 303) authenticates the user 304 through a cloud-based server to ensure that each session is linked to a specific user account for tracking progress and generating personalized feedback. Cloud-based server is a backend service that handles user data, authentication, and personalized settings for each session. The DFTA 303 retrieves user-specific data such as age, weight, height, and fitness goals. The DFTA 303 uses the user-specific data for exercise personalization and accurate calorie estimation. For example, for a user 304 whose fitness goal is weight loss, the DFTA 303 adjusts exercise intensity to match calorie-burning targets. After successfully completing the verification process, the DFTA 303 completes the initialization, optimizing the user device 200 for pose detection and real-time exercise tracking, thereby enabling smooth and accurate performance during the exercise session. Furthermore, the DFTA 303 performs real-time pose analysis and feedback generation locally on the user device to ensure low-latency operation, without involvement of the cloud-based server.

[0014] The dynamic fitness training assistant (DFTA 303) prompts 101 the user 304 to activate a front camera of the user device 200 to capture the live video 202 of the user 304. This real-time live video stream allows the users 304 to visualize their exercise performance on the display screen of the user device 200 as they perform an exercise. The live video 202 acts like a virtual mirror, allowing for precise posture correction and real-time feedback. To facilitate seamless motion tracking and enhance the accuracy of pose estimation, the DFTA 303 is configured to capture the live video 202 at a predetermined frame rate, adapted based on confidence in estimating lighting conditions. For example, when the lighting conditions are favorable, the DFTA 303 captures a live video 202 of a user 304 using a camera setting with a frame rate of 24 frames per second (FPS) for smooth, real-time motion tracking and accurate pose estimation. Additionally, the DFTA 303 ensures a minimum resolution of 1080 pixels (1080p) poser to enable accurate key point or landmarks detection. In an embodiment, although the live video is captured at a predetermined frame rate, posture evaluation and user feedback are decoupled from the frame rate and are not generated for every captured frame.

[0015] The dynamic fitness training assistant (DFTA 303) employs computer vision techniques, including a pose estimation model and a pose estimation algorithm, to analyze 102 the captured live video 202 and detects key points or landmarks of the user 304. For example, the pose estimation model comprises OpenPose, MediaPipe, MoveNet, etc., and pose estimation algorithm includes convolutional neural networks, keypoint detection, regression models, etc. As used herein, the term “key points” or “landmarks” refer to specific anatomical locations on the user's body, such as, joints and body parts, which are used to track their position and movement. The DFTA 303 does not record the captured live video. The DFTA 303 processes each frame of the captured live video 202 in bitmap format to ensure compatibility with its underlying machine-learning pose estimation algorithm. The pose estimation algorithm is configured to detect and extract key points of the user 304 from the bitmap format. For example, during exercise tracking, the pose estimation algorithm analyzes each video frame to detect key points on the user's body based on the type of the exercise. For example, the key points on the user's body comprise: 1) facial features: nose, left eye (inner, center, outer), right eye (inner, center, outer); 2) ears: left ear, right ear; 3) upper extremities: left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist; 4) pelvis and lower limbs: left hip, right hip, left knee, right knee, left ankle, right ankle, left heel, right heel, left foot (big toe area), right foot (big toe area). After detecting and extracting the key points of the user 304, the DFTA 303 assigns a confidence score ranging from 0 and 1 to each detected key point. For robust and accurate tracking, the confidence score of every detected key point must surpass the established threshold, thereby eliminating low-certainty detections from the analytical pipeline. Furthermore, when the confidence score of one or more required key points falls below the established threshold, downstream posture evaluation and feedback generation are temporarily suspended until stable confidence is restored.

[0016] The dynamic fitness training assistant (DFTA 303) verifies 103 orientation of the user device 200 and visibility conditions through guided interaction, and provides directional prompts to position the user 304 for optimal key point detection. Once the user 304 is properly positioned, the DFTA 303 overlays the optimal key points onto the analyzed frames of the live video 202 of the user 304 upon reaching a predefined detection confidence. To verify full-body visibility of the user 304, the DFTA 303 executes a key points based visibility check using the overlaid optimal key points. The DFTA 303 performs a visibility check by verifying that the confidence score of all the critical key points, for example, hips, knees, shoulders, etc., is greater than 0.7. If any single critical key point falls below the confidence score of 0.7, the DFTA 303 considers that the user 304 is not fully visible and issues an audio and / or visual prompts to adjust their position. For example, if a critical key point has a confidence score below 0.7, the DFTA 303 issues an audio prompt and / or a visual prompt “Make sure your full body is visible.”, every few seconds to guide the user 304 to adjust their position. The overlayed optimal key points onto the live video feed of the user 304 provide visual guidance on posture and alignment, enabling users 304 to adjust their position for accurate tracking. Once an audio or visual prompt is issued, the DFTA 303 defers subsequent visibility prompts until completion of the ongoing prompt and reassessment of the user's position.

[0017] Upon successful verification of full-body visibility of the user 304, the dynamic fitness training assistant (DFTA 303) calculates joint angles of the detected and extracted key points to monitor posture and exercise movement of the user 304. For example, during exercises like squats, the DFTA 303 calculates the hip-knee-ankle angle to determine squat depth. Similarly, for push-ups, the shoulder-elbow-wrist angle is used to ensure correct arm movement. In an embodiment, a normalized coordinate system is used for calculating angles. The normalized coordinate system outputs coordinates between 0 and 1, and these values can be scaled to any screen size. To ensure consistency, reference lines can be drawn at (Kx,0) for horizontal and (0, Ky) for vertical. The calculated joint angles are evaluated over a temporal window to assess movement consistency, rather than being acted upon based on instantaneous frame-level variations.

[0018] The normalized coordinate system defines spatial positions within a standardized range, facilitating accurate tracking and measurement. The system uses the following axes:

[0019] 1. X-axis: Ranges from 0 (left) to 1 (right).

[0020] 2. Y-axis: Ranges from 0 (top) to 1 (bottom).

[0021] 3. The origin point (0, 0) is located at the top-left of the screen, providing a consistent reference point for all calculations.

[0022] Furthermore, the normalized coordinate system employs reference points such as

[0023] (Kx,0) and (0,Ky) are used to facilitate tracking and measurement.

[0024] (Kx,0):(Kx, 0) reference point ensures that a reference line is horizontal at the same x-coordinate as the point, allowing for accurate tracking horizontal movements.

[0025] (0, Ky):(0, Ky) reference point provides a vertical reference line corresponding to the point's y-coordinate, for identifying vertical movements.

[0026] Examples of how angles are calculated are shown below:

[0027] Example 1. Calculation of Hip-Knee-Ankle Angle—The DFTA 303 calculates the angle at the Knee formed by the hip, knee, and ankle using normalized coordinates.Steps:1. Given Points:

[0029] Hip (Hx, Hy)

[0030] Knee (Kx, Ky)

[0031] Ankle (Ax, Ay)

[0032] 2. The given points Hip (Hx, Hy), Knee (Kx, Ky), and Ankle (Ax, Ay) are used to form two Vectors:

[0033] Vector 1 (Hip to Knee): v1=(Kx−Hx, Ky−Hy)

[0034] Vector 2 (Knee to Ankle): v2=(Ax−Kx, Ay−Ky)

[0035] 3. The angle θ between these two vectors is calculated using the dot product formula:cos (θ)=v1·v2 / ∥v1∥∥v2∥

[0036] The calculated cosine value is then converted into degrees to determine the Hip-Knee-Ankle Angle, which helps to evaluate movement quality and joint behavior during exercises such as squats.

[0037] Example 2. Hip-Knee-Horizontal (Relative to (Kx, 0))—The DFTA 303 calculates the angle between the Hip to Knee vector and a horizontal line starting at (Kx,0).Steps:1. Given Points:

[0039] Hip (Hx, Hy)

[0040] Knee (Kx, Ky)

[0041] Horizontal Reference: (Kx,0)

[0042] 2. The given points Hip (Hx, Hy), Knee (Kx, Ky), and Horizontal Reference: (Kx,0) are used to form two Vectors:

[0043] Vector 1 (Hip to Knee): v1=(Kx−Hx, Ky−Hy)

[0044] Vector 2 (Horizontal): v2=(0, Ky)

[0045] 3. The angle θ between these two vectors is calculated using the dot product formula:cos (θ)=v1·v2 / ∥v1∥∥v2∥

[0046] The resulting cosine value is then converted into degrees to determine the angle.

[0047] Example 3. Hip-Knee-Vertical (Relative to (0, Ky))—The DFTA 303 calculates the angle between the Hip to Knee vector and a vertical line starting at (0, Ky).Steps:1. Given Points:

[0049] Hip (Hx, Hy)

[0050] Knee (K, Ky)

[0051] Vertical Reference: (0, Ky)

[0052] 2. The given points Hip (Hx, Hy), Knee (Kx, Ky), and Vertical Reference: (0, Ky) are used to form two Vectors:

[0053] Vector 1 (Hip to Knee): v1=(Kx−Hx, Ky−Hy)

[0054] Vector 2 (Vertical): v2=(Kx,0)

[0055] 3. The angle θ between these two vectors is calculated using the dot product formula:cos (θ)=v1·v2 / ∥v1∥∥v2∥

[0056] The resulting cosine value is then converted into degrees to determine the angle.

[0057] The dynamic fitness training assistant (DFTA 303) ensures that full-body visibility is confirmed before initiating exercise tracking. To achieve this, the DFTA 303 employs a verification process that checks for the presence of all key points. If any critical key points are missing from view, the DFTA 303 will prompt the user 304 to adjust their position every few seconds until all key points or landmarks are detected. For example, if the user's feet are not fully visible in the frame, the DFTA 303 will issue an alert asking the user 304 to move back or adjust the camera angle to capture the entire body. In an embodiment, this process is repeated every few seconds until all key points are detected and confirmed, ensuring that the exercise tracking process can begin with accurate and reliable data. Following issuance of a visibility-related prompt, the DFTA 303 defers subsequent prompts until completion of the prompt and reassessment of the user's visibility state.

[0058] Once full-body visibility is ensured and angles are calculated, the dynamic fitness training assistant (DFTA 303) creates 104 a virtual workout box (VWB) around the user 304 in real-time based on an initial pose detected in the live video. The DFTA 303 maintains the created VWB as a persistent spatial constraint during performance of the exercise by anchoring the created VWB to the initial user pose, without recalculating the created VWB boundaries on a per-frame basis. The DFTA 303 uses the created VWB to define a region of interest for exercise tracking. To create the VWB, the DFTA 303 first performs a full-body visibility check to confirm that the full body of the user 304 is visible. Once full-body visibility is confirmed, the DFTA 303 evaluates whether joint angles are within the acceptable range for a selected exercise, and performs a joint angle analysis to ensure the user 304 is in the correct starting position. If joint angles match the predefined thresholds, the DFTA 303 proceeds to create the VWB for the exercise. For example, 1) Hip-Knee-Horizontal (HKH) angle is used to determine the depth and position for performing squats. 2) Shoulder-Elbow-Wrist angle is used for exercises like push-ups to monitor arm alignment. Depending on the type of exercise, the DFTA 303 creates a personalized VWB that takes into account specific movement patterns and joint angles. By creating a VWB in real-time, the DFTA 303 provides users 304 with a visual representation of their movements, enabling them to adjust their form and technique for optimal performance. Once created, the virtual workout box remains fixed for the selected exercise and is used as a persistent spatial constraint to enable or suspend exercise analysis, rather than being recalculated for each video frame.

[0059] After full-body visibility and joint angles are confirmed, the DFTA 303 extracts coordinates for the key points such as shoulders, hips, knees, and feet. Using the extracted key points data, the DFTA 303 computes the virtual workout box (VWB) corners based on user-specific anthropometric data. This ensures that the VWB precisely aligns with the user's body, allowing for accurate tracking. The VWB is defined by four corner points: 1) Top-left: (Xmin, Ymin); 2) Top-right: (Xmax, Ymin); 3) Bottom-left: (Xmin, Ymax); and 4) Bottom-right: (Xmax, Ymax). These corner points are determined based on the user's anthropometric measurements: 1) the corner points Xmin and Xmax are determined by the user's shoulder width and foot position; and the corner points Ymin and Ymax are determined based on the user's height and hip position. For example, when a user 304 is performing a squat exercise, the four corner points of dynamic VWB might be: 1) Xmin: Left foot X coordinate—margin; 2) Xmax: Right foot X coordinate+margin; 3) Ymin: Hip Y coordinate—upper margin; and 4) Ymax: Foot Y coordinate+lower margin. The dimensions of the VWB varies depending on the exercise being performed and serves as the region of interest for tracking user performance. For example, for push-ups, the VWB is formed at the starting position of the push-up. For example, for high knees, the VWB is formed when the user raises their hands. For example, for jumping jacks, the VWB is created when the user stands facing the camera with hands raised in a V-shape.

[0060] After the creation of the virtual workout box (VWB) around the user 304, the dynamic fitness training assistant (DFTA 303) defines a region of interest (ROI) that is used to analyze and track the user's movements. In an embodiment, the ROI is the VWB. The DFTA 303 pauses and resumes data capture and analysis based on the user's position relative to the created VWB and the front camera of the user device 200. For example, if the user 304 is within the VWB, the DFTA 303 continues tracking the exercise and provides real-time feedback. However, if the user 304 moves outside the VWB, the DFTA 303 pauses exercise tracking to prevent incorrect data collection. Furthermore, the VWB pauses and resumes data capture and analysis based on the user's position relative to the VWB. If the user 304 is inside the VWB, the DFTA 303 tracks the user's movement and collects data. If the user 304 moves outside the VWB, the DFTA 303 stops tracking and prompts the user 304, for example, “Move back into the workout box”, to return to the VWB. The DFTA 303 creates the VWB based on user-specific parameters, such as, posture, body structure, etc., and exercise specific configuration data. The created virtual workout box remains unchanged until completion of each exercise. The VWB acts as a boundary around the user's body, tracking movements and analyzing posture. The VWB adapts to each user's anthropometric data, such as height and limb length, ensuring accurate tracking for various exercises. When tracking is paused due to the user exiting the VWB, the DFTA 303 resumes analysis from the current exercise state upon user's re-entry in to the VWB rather than restarting the exercise or recalculating the VWB.

[0061] Before allowing the user 304 to begin the workout, the dynamic fitness training assistant (DFTA 303) determines 105 whether the user 304 is within the created virtual workout box (VWB) and in the correct starting position of the selected exercise. If the user's posture is incorrect, the DFTA 303 provides audio feedback, for example, “Adjust your stance.”, “Keep your back straight.”, etc., to guide them into the proper position. The DFTA 303 only starts tracking the exercise once the user's posture is corrected. The DFTA 303 also ensures that the user 304 stays within the VWB by providing guidance and prompts, for example, “Move back into the workout box”. Once the user's posture is correct, the DFTA 303 updates the text layout on the GUI to provide guidance, displaying a message, for example, “Virtual workout box is formed, start.”. Upon determining that the user 304 is within the VWB and is in the correct starting position, the DFTA 303 monitors and analyses the exercise movement of the user 304 using the detected and extracted key points and the calculated joint angles. As the user 304 performs the exercise, the DFTA 303 tracks 106 the exercise movement of the user 304 using the detected and extracted key points and the calculated joint angles to identify type of exercise performed by the user 304 and count the number of repetitions. The DFTA 303 continuously analyzes motion data to ensure precise motion capture. For example, the DFTA 303 begins tracking user's exercise movements frame-by-frame, for example, at 24 frames per second (FPS), to ensure precise motion capture. The DFTA 303 identifies multiple states while tracking the user's exercise movements, comprising: 1) rest state: the starting position with muscles relaxed; 2) active state: the user 304 is actively performing the exercise; 3) normal state: the user 304 is in a normal, non-exercise state; 4) trans state: a transitional state between rest and active states; and 5) pass state: a state where the user 304 is moving from one exercise to another. The number of states vary depending on the exercise being performed. For example, when performing squats if the Hip-Knee-Horizontal (HKH) angle is 0, the user is considered at rest. Although motion data is analyzed frame-by-frame, the DFTA 303 uses state transitions to regulate feedback generation so that corrective guidance is not issued continuously for every frame. Logic:if abs(current_hip_angle) < rest threshold: rest_state = Trueelse: rest_state = False

[0062] In an embodiment, if the user stays in the rest state for too long, for example, 5 seconds, the DFTA 303 provides an audio prompt to encourage the user 304 to stay engaged in the workout. For example, “Squats: Lower your hips”, “Stay engaged in the workout.”.

[0063] The active state occurs when the user 304 begins movement or is already performing the exercise. The DFTA 303 detects a transition from rest state to active state when there is a significant change in the calculated joint angles. Logic:if abs(current_hip_angle) > threshold: active_state = True

[0064] Once the user 304 is in the active state, the DFTA 303 analyzes the user's movements frame-by-frame using the real-time video input and detects key points, and calculates joint angles of the user in real-time. The DFTA 303 processes each individual frame of the live video feed sequentially, allowing the DFTA 303 to monitor and interpret changes in the user's posture, position, and joint angles. This frame-by-frame tracking enables the DFTA 303 to detect movement consistency and the proper range of motion of the user 304. The DFTA 303 analyses the detected key points and the calculated joint angles to identify the exercise performed by the user 304 and count the number of repetitions. The DFTA 303 counts the repetitions of the identified exercise being performed by tracking the user's exercise movement from a rest state to an active state to back to the rest state within a respective virtual workout box, which is the region of interest (ROI). The DFTA 303 counts a repetition as completed when it detects a full movement cycle, from a rest state to an active state and back to a rest state, of the identified exercise. For example, the logic used for counting the number of squat repetitions performed by the user 304 is shown below. The DFTA 303 determines correctness of an exercise repetition only upon completion of the full movement cycle, based on deviations accumulated during the active state, rather than being finalized on individual frames. if previous_state == ‘active’ and current_state == ‘rest’: rep_count += 1 if not is_incorrect:  correct_count += 1 else:  incorrect_count += 1 is_incorrect = Falseif current_state == ‘active’ and form_is_wrong: is_incorrect = Trueprevious_state = current_state

[0065] Each exercise has a predefined joint angle range that indicates correct movement. For example, the expected joint angle ranges for squats and push-ups are shown below. If the calculated joint angles fall within the expected range for the identified exercise, the DFTA 303 interprets the movement as correct and marks the repetition as valid. However, if calculated joint angles exceed the expected range of the identified exercise, the DFTA 303 flags the repetition as incorrect. The DFTA 303 calculates that the total number of correct repetitions performed by the user 304 as equal to the difference between the total number of repetitions performed by the user 304 minus the incorrect number of repetitions performed by the user 304. Formula: Total correct repetitions=Total repetitions−Incorrect repetitions.Example 1: Squat Angle RangesJointExpected Angle RangeHip-Knee0°-35° - RestVertical35°-50° - Incomplete>90° - Squat Too DeepBack150°-180° (Straight)Example 2: Push-Up Angle RangesJointExpected Angle RangeElbow90° (bottom), 180° (top)Shoulder~45° in descentBack170°-180° (Straight)The dynamic fitness training assistant (DFTA 303) identifies 107 deviations from predefined exercise motion pattern. The deviations detected during an active movement phase are accumulated and used to generate corrective feedback, and the accumulated deviations are evaluated at repetition completion boundaries using cumulative user movement data of the identified exercise to detect errors in the user's movement. The DFTA 303 analyses the recorded joint angles to identify any deviations that exceed the expected range. If the deviation is detected, the DFTA 303 flags the issue and generates a real-time feedback prompt to guide the user 304 back into the correct movement pattern. For example, if the user 304 is performing squats and the hip-knee angle is greater than 95 degrees, the DFTA 303 flags that the squat as too deep and generates a real-time feedback prompt, for example, “Incomplete squat! Lower your hips.”. This feedback helps the user 304 adjust their movement to ensure proper form and avoid injury. The detected deviations are accumulated during the active state and evaluated at repetition-level boundaries before triggering corrective feedback. Logic of a real-time feedback prompt:if hip_knee_angle > 95: print(“Squat too deep!”)Examples of different types of incorrect postures corresponding to different exercises is:Squats:

[0069] Incomplete→Hip-Knee angle too small) (<50°.

[0070] Too Deep→Hip-Knee angle too large (>) 95°.

[0071] Push-ups:

[0072] Elbows too wide→Elbow angle >100°.

[0073] Hips sagging→Back angle <170°.

[0074] Knees→Knee angle <150

[0075] Furthermore, the dynamic fitness training assistant (DFTA 303) identifies each incorrect repetition and categorizes them based on the type of error detected, as shown in the below table. The flagged errors are stored and presented to the user 304 in a summary report, providing a detailed breakdown of the types of mistakes made during the exercise. This feedback helps users 304 identify areas for improvement and adjust their movement patterns accordingly.Condition (Example:Mistake TypeSquats)Correction PromptIncomplete RepHip-Knee angle < 50°“Lower your hips!”Too Deep SquatHip-Knee angle > 95°“Don't go too deep!”Knees GoingKnee over toe“Keep knees behindForwardtoes!”

[0076] The dynamic fitness training assistant (DFTA 303) provides 108 real-time corrective media feedback to the user 304 based on the detected errors to assist the user 304 in maintaining proper form. Upon detecting incorrect movement or errors in the user movements, the DFTA 303 triggers one or more real-time feedback: audio, text or a combination of both. Examples of feedback prompts for different exercises include: 1) Squats-prompt: “Lower your hips!” 2) Push-Ups—prompt: “Keep your back straight!”; and 3) Lunges-prompt: “Don't let your knees go past your toes!”. In an embodiment, the DFTA 303 further comprises anomaly detection to identify incorrect form arising from improper execution or limited range of motion of the user 304 due to one of physical constraints, injuries, body structure anomalies, etc., of the user 304. In another embodiment, if the user 304 remains inactive or in a rest state for too long, the DFTA 303 generates an audio prompt, for example, “Continue your exercise!”, and prompts the user 304 to continue.

[0077] The dynamic fitness training assistant (DFTA 303) autonomously regulates energy targets in response to variations in exercise difficulty and user progression. This ensures that users 304 achieve a personalized fitness outcome based on their individual needs and goals. The DFTA 303 tracks the user's workout session by recording the start time at the beginning of the first repetition and the end time upon completion of the final repetition. Total workout duration is calculated as the difference between the end time and start time: Total workout duration=end time-start time. At the end of an exercise session, the dynamic fitness training assistant (DFTA 303) classifies 109 the user's exercise performance based on the tracked movements and the detected errors. The DFTA 303 grades user skill levels for each repetition based on angle and adapts feedback based on repetition accuracy, consistency, and speed of the user 304. For example, when performing squats, the DFTA 303 classifies the user's exercise performance into four categories:Performance LevelPerformance PercentageBeginner50-70Intermediate70-80Expert80-95ErrorSquat too deep

[0078] The dynamic fitness training assistant (DFTA 303) estimates 110 energy expenditure and dynamically maintains a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics. In an embodiment, the DFTA 303 comprises a calorie counter that adjusts calorie counts based on the user's performance, height, weight, and the number of repetitions performed. Errors for each repetition are displayed to provide users 304 with a clear understanding of their progress.

[0079] In an embodiment, at the end of the workout session, the DFTA 303 classifies the user's performance into a performance category based on their accuracy, consistency, and speed of the user 304. This classification provides users 304 with a comprehensive assessment of their exercise performance and helps them identify areas for improvement.Performance LevelPerformance scoreNovice<40Skilled user40 < x < 80Mastered user>80

[0080] For example, when a user 304 is performing squats, the DFTA 303 tracks the user's movement and counts each repetition and categorizes them based on their accuracy as: 1) repetitions 1 and 5 have beginner accuracy; 2) repetition 2 has intermediate accuracy; 3) repetitions 3 and 8 have expert accuracy; 4) repetitions 4, 6, 7 are too deep. The logic of performance calculation and classification is shown below:  output = [ 1, 5 repetitions beginner, 2 repetitions, intermediate 3, 8 repetitionsexpert, 4, 6, 7 repetitions too deep]score = { beginner_score : 14, intermeditae_score : 16, expert_score : 20, squattoo deep : −10}performance = 2*14 + 1*16 + 2*20 + 3 *−10 = 54# classified as Skilled

[0081] Upon completion of each exercise, the dynamic fitness training assistant (DFTA 303) generates an interactive post-exercise analysis interface configured to present performance insights and feedback indicators reflective of user activity, accuracy, and progression. In an embodiment, after tracking the user's exercise session in real-time, the DFTA 303 generates a detailed summary report to evaluate the user's exercise / workout session by summarizing key metrics: 1) total duration of the workout; 2) correct and incorrect number of repetitions; 3) calories burned during each exercise session; 4) performance classification (novice, skilled, mastered); and 5) recommendations for improvement. The generated summary report enables users 304 to understand their strengths, identify errors, and track progress over time. In an embodiment, the summary report of the user's exercise session includes: 1) Total Workout Duration→The amount of time the user 304 spent performing the exercise; 2) Total Repetitions Counted→There is a predefined number of repetitions that users 304 perform per set. The number of repetitions completed (correct+ incorrect number of repetitions); 3) Correct Repetitions→The number of repetitions performed with the correct form; 4) Incorrect Repetitions→The number of repetitions flagged due to errors in form or depth; 5) Calories Burned→An estimate based on user data and exercise intensity; and 6) Performance Level Classification→Novice or Skilled or Mastered. Example of a summary report of the user's exercise session:

[0082] Workout Summary:

[0083] Total Duration: 10 minutes

[0084] Total Reps: 15

[0085] Correct Reps: 10

[0086] Incorrect Reps: 5

[0087] Calories Burned: 125 kcal

[0088] Performance Level: Skilled

[0089] Adjustments Recommended: Improve knee positioning in squats.

[0090] Summary: Repetitions 1 and 5 have beginner accuracy, repetition 2 has intermediate accuracy, repetitions 3 and 8 have expert accuracy, and repetitions 4, 6, and 7 are too deep.

[0091] Second example of a summary report of the user's exercise session:

[0092] Workout Summary:

[0093] Total Duration: 10 minutes

[0094] Total Reps: 15

[0095] Correct Reps: 10

[0096] Incorrect Reps: 5

[0097] Performance Level: Skilled

[0098] Adjustments Recommended (based on repetitive mistakes): Keep knees aligned in squats.

[0099] Suggested Next Level: Maintain current exercise for 2 more sessions before progressing.

[0100] The dynamic fitness training assistant (DFTA 303) tracks the user's performance over a predefined duration and adjusts the difficulty of the user's future workouts based on their performance level. If the user 304 performs consistently well over a predefined duration, the DFTA 303 assigns harder exercises in future workouts. For example, if the user 304 achieves a rating of Mastered Level for multiple workout sessions, the DFTA 303 assigns a harder exercise for future workouts. The DFTA 303 ensures that the total number of calories burned by the user / session remains consistent despite increased intensity. Conversely, if the user 304 performs consistently poorly over a predefined duration or if the workout errors persist, the DFTA 303 assigns easier variations of the exercises in future workouts. Furthermore, the DFTA 303 is configured to maintain calorie balance even if the exercise difficulty is adjusted. For example, if the user 304 remains at the beginner level for too long or multiple workout sessions, the DFTA 303 assigns an easier variation of the exercises for future workouts to prevent plateaus and maintain caloric goals. The DFTA 303 ensures caloric goals stay balanced even with the easier variation. In an embodiment, the DFTA 303 tracks user's general joint angles for various exercises over a predefined duration, such as on a monthly basis. Based on the detected user's performance trends and deviation in joint angles, the DFTA 303 detects anomalies due to one of physical constraints, injuries, body structure anomalies, etc., of the user 304 and automatically adjusts the user's exercise thresholds and calculates future performance accordingly. Example of the DFTA 303 decision flow is shown below:if level == “Mastered” and consecutive expert sessions >= 3: next exercise difficulty += 1 # Progress to harder exerciseelif level == “Novice”: next_exercise_difficulty -= 1 # Adjust to easier variationelse: next exercise difficulty = current difficulty # Maintain current level

[0101] In an embodiment, the dynamic fitness training assistant (DFTA 303) is configured to provide users 304 with a comprehensive report of their performance across multiple workout sessions. The comprehensive report helps users 304 to track their progress, identify trends, detect recurring mistakes, and receive personalized workout adjustments based on historical performance. For example, at the end of a workout week, the DFTA 303 compiles workout / exercise session data and generates a progress report that includes:

[0102] 1. Total Workouts Completed→Number of workout sessions performed this week.

[0103] 2. Best Workout Session→The workout session with the highest accuracy.

[0104] 3. Performance Trend Over Time→Tracking improvement in correct repetitions vs. incorrect repetitions.

[0105] 4. Common Mistakes Identified→Frequent errors were detected across workout sessions.

[0106] 5. Workout Adaptation Plan→Recommendations for next week's exercises.

[0107] The dynamic fitness training assistant (DFTA 303) determines whether to increase, maintain, or decrease exercise difficulty of future exercise sessions, by evaluating: 1) Consistency: The DFTA 303 measures how stable the user's performance is across all exercise sessions. If the user 304 performs consistently well, the DFTA 303 determines that the user 304 may be ready for a more advanced workout; 2) Error Rate is defined as the percentage of incorrect repetitions over a total number of repetitions. If the error rate remains high, the DFTA 303 suggests adjustments to correct the user's form / posture before progressing; 3) Total Correct Repetitions: the DFTA 303 uses the total number of correct repetitions to assess whether the user 304 can handle a more challenging workout; and 4) Performance Classification Trends: the DFTA 303 tracks whether the user 304 is progressing from Novice→Skilled→Mastered over time.

[0108] An example of performance thresholds used by the dynamic fitness training assistant (DFTA 303) for an adjustment workout is shown in the table below.Performance Level% of Incorrect RepsAction TakenNovice>40% IncorrectAssign an easier variationSkilled10%-40% IncorrectMaintain current workoutMastered<10% IncorrectIncrease difficulty

[0109] The dynamic fitness training assistant (DFTA 303) aggregates data from all workout sessions of the week and generates a comprehensive summary report. An example of a weekly performance report output is shown below:

[0110] Weekly Performance Report:

[0111] Total Workouts: 5

[0112] Best Session: Session 4 (Accuracy: 95%)

[0113] Common Mistakes: Knees too forward in squats

[0114] Total Correct Reps: 72

[0115] Total Incorrect Reps: 18

[0116] Performance Trend: Improving (Novice→Skilled)

[0117] Suggested Progression: Move to Push-ups next session

[0118] In an embodiment, the dynamic fitness training assistant (DFTA 303) generates a progress chart showing accuracy over multiple sessions. The progress chart helps users 304 to understand their improvements or lack of improvements over multiple sessions. An example bar chart representation of weekly progress of a user 304 is shown in the table below:SessionCorrect RepsIncorrect RepsSession 1105Session 2124Session 3143Session 4161Session 5170

[0119] Key Insights a user can infer from the above bar chart: 1) The user's accuracy across the session is increasing, as indicated by the fewer number of incorrect repetitions: 2) Session 5 has zero mistakes, meaning that the user 304 is ready for progression to a more advanced level i.e., increase exercise difficulty; and 3) Common mistake detection (e.g., knee positioning in squats) suggests targeted corrections before advancing.

[0120] In another embodiment, the dynamic fitness training assistant (DFTA 303) is configured to generate an adaptive workout plan for future workouts based on weekly trends. For example, the DFTA 303 customizes the next week's workout plan based on the past week's trend. If the user 304 has consistently achieved the expert level rating, the DFTA 303 assigns the next difficulty level. For example, progress from kneeling push-ups→push-ups. If the user 304 is struggling or consistently achieves a beginner level rating for multiple sessions, the DFTA 303 assigns easier variations to focus on proper form and technique. Example of DFTA decision logic:· if incorrect_reps / total_reps > 0.4:· next_exercise_difficulty -= 1 # Assign easier variation· elif incorrect_reps / total_reps < 0.1:· next_exercise_difficulty += 1 # Assign harder variation· else:· next_exercise_difficulty = current_difficulty # Maintain current level

[0121] The dynamic fitness training assistant (DFTA 303) is further configured to provide personalized recommendations for the user's next workout cycle based on weekly summary or progress report. Example adjustments based on summary / progress report:Current ExerciseRecommended ChangeWall PushupsInclined PushupsInclined PushupsKneeling PushupsKneeling PushupsPushups

[0122] At the end of the week, the system generates a summary message with actionable insights. Example of a weekly summary message:

[0123] Weekly Performance Report:

[0124] Sessions Completed: 5

[0125] Best Session: Session 4 (95% accuracy)

[0126] Common Mistakes: Elbow position is incorrect

[0127] Average Error Rate: 10% (Skilled Level)

[0128] Next Steps: Progress to Push-ups

[0129] Adjustments Recommended: Focus on balance and elbow alignment

[0130] In an embodiment, the dynamic fitness training assistant (DFTA 303) is configured to perform long-term progress tracking. The DFTA 303 analyzes the user's fitness journey over multiple weeks or months by: 1) tracking historical performance trends; 2) detecting long-term improvements; and 3) providing personalized progression plans based on data-driven insights. The long-term progress tracking ensures that users 304 stay motivated and continue improving, as they can see their progress over time and make adjustments accordingly. Example of DFTA decision logic for long-term progress tracking: 1) If user 304 shows consistent improvement over 4+ weeks, recommend increasing exercise intensity by 1 level; 2) If user 304 achieves a plateau in performance after 2+ months, suggest adjusting workout routine to include more variety or increasing exercise frequency; and 3) If user 304 shows significant regression in performance over 3+ months, recommend taking a rest period and revisiting workout goals.

[0131] Unlike weekly reports, which focus on short-term progress, in long-term tracking, the DFTA 303 aggregates data across multiple weeks / months to:

[0132] Identify performance trends of the user 304, as indicated by improvement, stagnation, or decline over time.

[0133] Provide historical comparisons of accuracy and workout consistency.

[0134] Adjust training intensity, volume, and variation based on the user's progress.

[0135] Detect anomalies or plateaus faced by the user 304, ensuring continuous improvement.

[0136] Generate personalized, long-term exercise plans based on the aggregated data.

[0137] Metrics Used for Long-Term Progress Evaluation: To effectively track long-term fitness improvements, the dynamic fitness training assistant (DFTA 303) considers:

[0138] 1 Performance Trends Over Time

[0139] Tracks the user's accuracy rate over weeks / months to determine improvement.

[0140] Measures changes in correct repetitions vs. incorrect repetitions over time.

[0141] 2 Workout Consistency

[0142] Evaluates how often the user 304 works out per week / month.

[0143] If workout frequency drops, the DFTA 303 sends motivational reminders to the user 304 to encourage consistency.

[0144] 3 Strength and Endurance Progression

[0145] The DFTA 303 tracks resistance levels, duration, and reps over time to measure strength gains.

[0146] If the user handles higher resistance easily, the DFTA 303 recommends intensity increases.

[0147] 4 Error Reduction Rate

[0148] The DFTA 303 measures how effectively the user 304 is correcting form / posture mistakes over multiple sessions.

[0149] Example: If knee misalignment was an issue in Week 1 but improved by Week 4, the system marks this as a corrected error.

[0150] 5 Exercise Progression and Adaptation.

[0151] The DFTA 303 tracks which exercises the user 304 has mastered and when to introduce new challenges.

[0152] Example: If a user 304 has performed Kneeling Push-ups for 4 weeks with 95% accuracy, the system suggests transitioning to Push-ups.

[0153] The dynamic fitness training assistant (DFTA 303) stores session data in a database or a comma-separated values (CSV) file for reference and to analyze long-term trends. The stored data includes: workout session date, exercise type and difficulty level, total repetitions (correct and incorrect), performance level (beginner, intermediate, expert), calories burned, and recommended exercise adjustments. The stored data is used to generate monthly reports that provide a broader view of the user's progress.

[0154] The dynamic fitness training assistant (DFTA 303) uses the stored data to generate trend graphs and charts that help users 304 visualize their long-term improvements. Example of Accuracy Progression Chart (correct repetitions vs. incorrect repetitions over time) is shown the table below. From the table, it can be inferred that the user 304 has steadily improved accuracy and reduced incorrect repetitions (reps) over four weeks.WeekCorrect RepsIncorrect RepsImprovement %Week 15020—Week 26015+28.57%Week 37510+33.24%Week 4855+19.53%

[0155] The improvement percentage is calculated based on the “volume of the exercise” and the “percentage of the correct repetitions”. The formula for calculating the improvement percentage is shown below:Overall Improvement (%)=[((New Correct Reps−Old Correct Reps) / Old Correct Reps)×100]+(Latest Correct %−Previous Correct %)

[0156] An example of a graph showing a user's transition from Novice→Skilled→Mastered. The graph shows the user's performance level distribution over time and using the graph, the DFTA identifies when users are ready for more advanced exercises.

[0157] Week 1-2: Novice Level (Performance percentage >40%)

[0158] Week 3-4: Skilled Level ((Performance percentage 40%-50%)

[0159] Week 5+: Mastered Level ((Performance percentage >80%)

[0160] The dynamic fitness training assistant (DFTA 303) categorizes users into different progress levels based on the stored historical data. The DFTA 303 generates a Long-Term performance classification table as shown below:Long-TermProgressLevelCriteriaActionNoviceWorkout consistency < 2x perEncourage more frequentweektrainingProgressingConsistency 3x per week,Maintain steadyaccuracy improvingprogressionAdvancedConsistently high accuracyIntroduce more advanced(95%+)exercisesPlateauingNo improvement for 3+ weeksAdjust workout intensity& variety

[0161] For example, if a user 304 is plateauing, the DFTA 303 suggests the user 304:1) Change workout variation (e.g., different squat types); 2) Adjust resistance or reps to increase challenge; and 3) Include cross-training (e.g., combining squats with lunges).

[0162] At the end of each month, the DFTA 303 generates a comprehensive progress report detailing achievements and future goals. An example monthly progress report output:**Monthly Fitness Progress Report**Total Workouts: 16

[0164] Best Session: Week 3 (Accuracy: 98%)

[0165] Avg. Error Rate: 12% (Skilled Level)

[0166] Most Common Mistake: Elbow alignment in Kneeling push-ups

[0167] Workout Progression: Transition from Kneeling Push-ups→Push-ups

[0168] Adjustments for Next Month: Increase rep count and add resistance

[0169] The DFTA 303 analyses long-term performance and generates a customized workout strategy or a future workout plan for the next month. If the accuracy of the user 304 remains above 90% for multiple sessions, the DFTA 303 increases the difficulty by:

[0170] Adding resistance (e.g., Wall Push-ups to Kneeling Push-ups).

[0171] Increasing reps or sets to build endurance.

[0172] Introducing progressive overload techniques for strength gain.

[0173] If performance of the user 304 declines or user fatigue is detected, the DFTA 303 suggests recovery-based modifications such as:

[0174] Lower-intensity exercises (e.g., switching from Kneeling push-ups to wall push-ups).

[0175] Reducing training frequency to allow muscle recovery.

[0176] Incorporating stretching or mobility work to prevent injuries.

[0177] The DFTA 303 analyzes weekly and monthly reports and creates a dynamic artificial intelligence (AI)-driven workout plan that adapts based on user performance. Furthermore, the AI monitors each session and dynamically adjusts workouts as the user 304 progresses. Example of an AI-generated workout adjustments based on user's long term progress is shown below:CurrentNextExerciseProgressionWall PushupsKneelingpushupsKneelingPushupspushups

[0178] FIGS. 2A-2L exemplarily illustrates screen shots of an embodiment of a dynamic fitness training assistant (DFTA 303) as displayed on a user device 200 during the real-time exercise tutoring. For the purposes of this example, it is understood that the DFTA 303 is installed on the user's device 200, for example, a smartphone, and the user 304 has completed the registration process with the DFTA 303. Furthermore, the smartphone is configured to facilitate real-time exercise tutoring as described earlier in this application. For the purposes of this example, the user 304 intends to perform the squats exercise. Before starting the workout exercise, the user 304 positions the user device 200 on the floor substantially perpendicular to the ground. Once the user device 200 is properly positioned, the user 304 starts the workout session by clicking start button provided by the DFTA 303. The DFTA 303 activates the front camera of the user device 200, captures a live video 202 of the user 304, provides an audio queue, displays visual instructions “Make your full body visible”203, and text layout counters for correct 204 and incorrect 205 repetitions on the graphical user interface 201 of the user device 200, as exemplarily illustrated in FIG. 2A.

[0179] The dynamic fitness training assistant (DFTA 303) analyses the captured live video 202 and detects key points 206 of the user 304 using a pose estimation model and a pose estimation algorithm, as exemplarily illustrated in FIG. 2B. The DFTA 303 provides audio instructions (not shown) and visual instructions “Make your full body visible”203 to ensure the full body of the user 304 is visible. If the user 304 is not facing the camera, the DFTA 303 instructs the user 304 to face the camera, as exemplarily illustrated in FIG. 2C. Upon successful verification of full-body visibility of the user 304, the DFTA 303 overlays the detected key points 206 onto the live video feed 202 of the user 304 to provide visual guidance on posture and alignment. After overlaying the detected key points 206 onto the live video feed 202 of the user 304, the DFTA 303 calculates joint angles of the detected key points 206 to monitor posture and exercise movement of the user 304 as described earlier in this application.

[0180] Once the user 304 is facing the camera, full-body visibility is ensured and angles are calculated, the dynamic fitness training assistant (DFTA 303) instructs the user 304 to ‘raise their hands sideways’207, as exemplarily illustrated in FIG. 2D. After the user 304 raises their hands sideways, the DFTA 303 creates a virtual workout box (VWB) 209 around the user 304 in real-time based on the detected key points 206, and defines a region of interest for exercise tracking, as exemplarily illustrated in FIG. 2E. The creation of the VWB is described earlier in this application. The DFTA displays visual instructions / prompts / guidance 208 required for performing squats exercise. Before the user 304 begins the workout, the DFTA 303 determines whether the user 304 is within the dynamic VWB 209 and the user 304 is in correct starting position of the squats exercise. If the user 304 moves outside the VWB 209, the VWB 209 and workout box turns red (not shown) and the DFTA 303 highlights the displayed visual prompt 208“outside the workout box”, as exemplarily illustrated in FIG. 2F. Furthermore, for every three seconds, the DFTA 303 plays an audio prompt “Go inside the virtual workout box.” The DFTA 303 does not start exercise tracking until the user's posture is corrected. Upon issuing an audio prompt, the DFTA 303 defers subsequent prompts until completion of the ongoing audio and reassessment of the user's position within the virtual workout box.

[0181] Upon determining that the user 304 is within the VWB 209 and is in correct starting position, the DFTA 303 instructs the user 304 to perform the squats exercise via an audio prompt: “Lower your hip,” and the displayed visual instruction “Lower your hip”208 turns green / highlighted, as exemplarily illustrated in FIG. 2G. The DFTA 303 monitors and analyses the squats exercise movement of the user 304 using the detected key points 206 and calculated joint angles, as exemplarily illustrated in FIG. 2H. The DFTA 303 begins tracking user's exercise movements frame-by-frame to ensure precise motion capture. The DFTA 303 identifies two or more states, for example, rest state, active state, normal state, trans state, and pass state, while tracking user's exercise movements. The rest state occurs when the user 304 is in the starting position with muscles relaxed. When the user 304 enters an active state, the DFTA 303 evaluates the repetition level using a predefined threshold, classifying it as, for example, Beginner, Intermediate, or Expert.

[0182] After the user 304 completes a single repetition of the squats exercise, the DFTA 303 checks if there are any errors in the active state. If there are no errors in the active state the DFTA 303 increments the correct repetition counter 204 and the count is announced via audio, as exemplarily illustrated in FIG. 2I. If there is any error in the active state, for example, if the user's squat position is too deep, the DFTA 303 provides / plays an audio prompt / cue: “Squat too deep”, and the displayed visual instruction 208“Squat too deep” turns red / highlighted, as exemplarily illustrated in FIG. 2J. After the user 304 completes repetition of the squats exercise, the DFTA 303 increments the incorrect repetition counter 205 as there is an error squat too deep in the active state, and the count is announced via audio, as exemplarily illustrated in FIG. 2K. Once the desired number of repetitions are completed, the DFTA 303 determines that the exercise activity ended, and generates a summary page, displaying a bar graph of correct and incorrect counts, the exercise duration, calories burned per session, and a bar graph showing the level classification, for example, Beginner, Intermediate, Expert, for each repetition, as exemplarily illustrated in FIG. 2L.

[0183] FIG. 3 illustrates an example of a dynamic fitness training assistant (DFTA) 303 implemented on a user device 200. The user device 200 is configured to execute real-time exercise tutoring. The user device 200 includes at least one processor 301 and a memory unit 302 for storing instructions. The memory unit 302 stores the DFTA 303. Th DFTA 303 comprises modules 303a-303f, each defining computer program instructions executable by the processor 301. The computer program instructions may perform additional steps, including those associated with real-time exercise tutoring as described in the specification. A single program or multiple programs may execute all or part of the method. The executable computer program instructions may be written in any suitable programming language and stored in any non-transitory computer-readable medium. The non-transitory computer-readable media may include centralized or distributed databases, server-based storage, cache systems, or any storage structure capable of encoding instructions for execution by a processor. Although FIG. 3 illustrates the DFTA 303 executing locally within the user device 200, the modules may alternatively execute remotely using a web browser, remote server, mobile device, distributed architecture, or combinations thereof. Portions of the modules 303a-303f may therefore be implemented across multiple computing devices connected via a network. The network may include any combination of wired or wireless communication technologies such as the Internet, mobile communication networks (e.g., 3G, 4G, 5G, LTE), Wi-Fi, satellite networks, local area networks, wide area networks, and other communication infrastructures or combinations thereof.

[0184] The memory unit 302 stores the modules 303a-303f of the DFTA 303, which may be implemented in software, hardware, firmware, or any combination thereof. The processor 301 is operably coupled to the memory unit 302 to execute the instructions of the DFTA 303. The memory unit 302 may include random-access memory (RAM), read-only memory (ROM), hard drives, solid-state drives, optical or magnetic media, or other non-transitory computer-readable storage media configured to store executable instructions, temporary variables, and runtime data. The processor 301 refers to one or more microprocessors, central processing unit (CPU) devices, finite state machines, computers, microcontrollers, digital signal processors, logic, a logic device, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a chip, etc., or any combination thereof, capable of executing computer programs or a series of commands, instructions, or state transitions. In an embodiment, the processor 301 is implemented as a processor set comprising, for example, a programmed microprocessor and a math or graphics co-processor. The DFTA 303 is not limited to employing the processor 301. In an embodiment, the DFTA 303 employs a controller or a microcontroller.

[0185] The modules 303a-303f of the DFTA 303 define instructions that, when executed by the processor 301, configure the user device 200 to implement real-time exercise tutoring. The DFTA 303 includes user interface 303a, processing module 303b, verification module 303c, bounding box module 303d, exercise module 303e, and notification module 303f. The processor 301 retrieves the instructions defined by the modules 303a-303f and executes their corresponding functions. The modules may be implemented as software executed by the processor, as hardware logic, as firmware, or as a combination thereof. As used herein, a “module” or “engine” refers to any combination of hardware, software, and / or firmware configured to perform the associated functions.

[0186] Upon activation, the DFTA 303 automatically configures the necessary components on a user device 200 to facilitate real-time exercise tutoring, as disclosed earlier in detailed description of the invention. The DFTA 303 is configured to prompt a user 304, via a graphical user interface (GUI) 201, to activate a front camera of the user device 200 to capture a live video 202 of the user 304, and allow the user 304 to visualize an exercise performance on the GUI 201 of the user device 200 as the user 304 performs an exercise.

[0187] The processing module 303b is configured to analyze the captured live video 202 of the user 304 and detect and extract key points of the user 304 using a pose estimation model and a pose estimation algorithm of the DFTA 303. The processing module 303b of the DFTA 303 analyses each frame of the captured live video 202 using computer vision techniques and machine learning algorithms to detect and track key points, as disclosed earlier in the detailed description of the invention.

[0188] The verification module 303c is configured to verify orientation and visibility conditions through guided interaction, and provides directional prompts to position the user 304 for optimal key point detection. The verification module 303c is further configured to overlay the optimal key points onto the analyzed frames of the live video 202 of the user 304 upon reaching a predefined detection confidence. Thereafter, the verification module 303c is further configured to verify the full-body visibility of the user 304 by executing a key point-based visibility check based on the overlaid optimal key points, as disclosed earlier in the detailed description of the invention. Upon successful verification, the DFTA 303 calculates joint angles of the detected and extracted key points to monitor posture and exercise movement of the user 304, as disclosed earlier in the detailed description of the invention.

[0189] Once full-body visibility is ensured and angles are calculated, the bounding box module 303d is configured to create a virtual workout box (VWB) around the user 304 based on an initial pose detected in the live video 202 in real-time. The bounding box module 303d is further configured to maintain the created VWB as a persistent spatial constraint during performance of the exercise by anchoring the created VWB to the initial user pose, without recalculating the created VWB boundaries on a per-frame basis. The created virtual workout box defines a region of interest for exercise tracking. The bounding box module 303d of the DFTA 303 creates the virtual workout box according to user-specific parameters and exercise-specific configuration data, and maintains the created virtual workout box unchanged until completion of each exercise, as disclosed earlier in the detailed description of the invention. Before allowing the user 304 to begin the workout, the exercise module 303e is configured to determine whether the user 304 is within the created virtual workout box and in the correct starting position of the selected exercise. Upon determining that the live video 202 of the user 304 is within the VWB and is in the correct starting position, the exercise module 303e is configured to monitor and analyze the exercise movement of the user 304 using the detected and extracted key points and the calculated joint angles. The exercise module 303e of the DFTA 303 is further configured to pause and resume data analysis based on user position relative to the created virtual workout box and the front camera of the user device 200. The virtual workout box is not recalculated for each video frame and is maintained as a persistent spatial constraint for the selected exercise once the starting posture conditions are satisfied.

[0190] In an embodiment, based on the detected key points and exercise-specific conditions, the DFTA 303 creates a virtual workout box defining a region of interest for exercise tracking. The DFTA 303 monitors user movement using detected key points and calculated joint angles, and selectively pauses and resumes data analysis based on the user's position relative to the virtual workout box.

[0191] As the user 304 performs the exercise, the exercise module 303e module is configured to track the exercise movement of the user 304 using the detected and extracted key points and the calculated joint angles to identify type of exercise performed by the user 304 and count number of repetitions. The exercise module 303e of the DFTA 303 continuously analyzes motion data to ensure precise motion capture, as earlier disclosed in detailed description of the invention. The exercise module 303e module is further configured to identify deviations from a predefined exercise motion pattern. The exercise module 303e module is further configured to accumulate the deviations detected during an active movement phase and use it to generate corrective feedback, and the accumulated deviations are evaluated at repetition completion boundaries using cumulative user movement data of the identified exercise to detect errors in the exercise movement of the user 304, as disclosed earlier in the detailed description of the invention. The exercise module 303e of the DFTA 303 further identifies incorrect form arising from improper execution or limited range of motion due to physical constraints or body structure anomalies. The notification module 303f module of the DFTA 303 is configured to provide a real-time corrective media feedback to the user 304 based on the detected errors to assist the user 304 in maintaining proper form, as earlier disclosed in the detailed description of the invention. The notification module 303f generates corrective media feedback based on deviations accumulated during an active movement phase and resolved at repetition boundaries, rather than reacting to instantaneous frame-level deviations.

[0192] The exercise module 303e module is further configured to classify the exercise performance of the user 304 based on the tracked movements and the detected errors. The exercise module 303e of the DFTA 303 grades user skill levels and adapts feedback based on repetition accuracy, consistency, and speed of the user, as earlier disclosed in the detailed description of the invention. The exercise module 303e module is further configured to estimate energy expenditure and dynamically maintaining a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics. The exercise module 303e of the DFTA 303 autonomously regulates energy targets in response to variations in exercise difficulty and user progression to achieve a personalized fitness outcome.

[0193] For example, the user 304 decides to perform squats using the user device 200. The user starts the DFTA 303 on the user device 200 and positions the user device 200 on a stable surface substantially perpendicular to the ground to achieve an optimal field of view for motion tracking. The user 304 begins the exercise by standing with his feet shoulder-width apart and his hands by his sides. The DFTA 303 captures live video 202 of the user movement using a front-facing camera of the user device 200. The DFTA 303 analyses each frame of the captured live video 202 and detects and extracts key points on the user's body, including the hips, knees, and ankles, using a pose estimation model and a pose estimation algorithm, such as, machine learning algorithms, for example convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The pose estimation model of the DFTA 303 processes each frame of the captured live video 202 by mapping pixel-level features to predefined anatomical key points using a trained neural network, outputting spatial coordinates and confidence scores for each detected joint. The DFTA 303 uses the pose estimation algorithm to identify the location and orientation of the user's body in each frame of the captured live video 202.

[0194] Once the key points have been detected, the DFTA 303 verifies orientation and visibility conditions through guided interaction to ensure accurate detection of the key points. The DFTA 303 uses the machine learning algorithm to analyze the user's movements and provides directional prompts to the user 304 to position himself correctly for optimal key point detection. The DFTA 303 employs AI-powered computer vision to detect and track key points on the user's body, including his hips, knees, and ankles. The machine learning model of the DFTA 303 evaluates spatial relationships and confidence scores of detected key points across successive frames to determine positional misalignment and generates corresponding directional prompts to guide the user into an optimal capture posture. For example, the DFTA 303 may prompt the user to “step forward with their left foot” or “bend their knees slightly”. After verifying the orientation and visibility conditions, the DFTA 303 creates a virtual workout box around the user 304 in real-time to define a region of interest for exercise tracking. The virtual workout box is generated by computing spatial boundaries from detected key points and anthropometric parameters once the user satisfies starting posture conditions, and is maintained as a persistent region of interest for the selected exercise. The DFTA 303 creates a bounding box that tracks the user's movements and adjusts its size and position based on the detected key points.

[0195] Next, the DFTA 303 monitors and analyzes the exercise movement of the user 304 using detected and extracted key points and calculated joint angles. The DFTA 303 tracks the movement of the user's body as he performs squats, including movement of the hips, knees, and ankles. The DFTA 303 analyzes temporal changes in joint angles derived from the detected key points across successive frames to assess movement consistency, range of motion, and posture alignment during the exercise. The DFTA 303 may also use the machine learning algorithms to identify patterns and relationships between different exercises and provide personalized feedback. As the user 304 performs squats, the DFTA 303 detects errors in his form and provides real-time corrective media feedback to assist him in maintaining proper form. The DFTA 303 generates audio cues, visual alerts, or tactile vibrations that are triggered by the system's detection of errors in the user's exercise movement. Upon identifying an error condition based on accumulated movement deviations, the DFTA 303 selects an appropriate feedback modality and triggers the corresponding audio, visual, or tactile cue to guide corrective action. For example, if the user 304 is not keeping his back straight or is allowing his knees to extend past his toes, the DFTA 303 may provide an audio cue saying “keep your back straight” or “bend your knees slightly more”. Finally, the DFTA 303 estimates energy expenditure and dynamically maintains a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics. The DFTA 303 analyzes data on the user's exercise performance and adjusts its energy expenditure calculations accordingly. The DFTA 303 estimates energy expenditure by combining user-specific metabolic parameters with exercise duration, repetition count, and movement intensity derived from joint-angle dynamics, and updates calorie calculations accordingly.

[0196] The non-transitory, computer-readable storage medium, herein exemplarily referred to as the memory unit 302, stores computer program instructions of a dynamic fitness training assistant (DFTA) 302 executable by at least one processor 301 for executing real-time exercise tutoring. The computer program instructions when executed by the processor 301 cause the at least one processor to: prompt a user 304 to activate a front camera of a user device 200 to capture a live video 202 of the user 304, and allow the user 304 to visualize an exercise performance on a GUI 201 as the user 304 performs an exercise; analyze the captured live video 202 of the user 304 using a pose estimation model and a pose estimation algorithm of DFTA 303 to detect and extract key points of the user 304, wherein the DFTA 303 analyses each frame of the captured live video 202 of the user 304; verify orientation and visibility conditions through guided interaction, provide directional prompts to position the user 304 for optimal key point detection, overlay the optimal key points onto the analyzed frames of the live video 202 of the user 304 upon reaching a predefined detection confidence, and thereafter verify the full-body visibility of the user 304 by executing a key point-based visibility check based on the overlaid optimal key points, wherein, upon successful verification, the DFTA 303 calculates joint angles of the detected and extracted key points to monitor posture and exercise movement of the user 304; create a virtual workout box around the user 304 based on an initial pose detected in the live video and maintain the created virtual workout box as a persistent spatial constraint during performance of the exercise by anchoring the created virtual workout box to the initial user pose, without recalculating the created virtual workout box boundaries on a per-frame basis, wherein the created virtual workout box defines a region of interest for exercise tracking, wherein the DFTA 303 creates the virtual workout box according to user-specific parameters and exercise-specific configuration data, and maintains the created virtual workout box unchanged until completion of each exercise; determine whether the user 304 is within the created virtual workout box, monitor and analyze the exercise movement of the user 304 using the detected and extracted key points and the calculated joint angles, wherein the DFTA 303 pauses and resumes data analysis based on user position relative to the created virtual workout box and the front camera of the user device 200; track the exercise movement of the user 304 using the detected and extracted key points and the calculated joint angles to identify type of exercise performed by the user 304 and count number of repetitions, wherein the DFTA 303 continuously analyzes motion data to ensure precise motion capture; identify deviations from a predefined exercise motion pattern, wherein the deviations detected during an active movement phase are accumulated and used to generate corrective feedback, and the accumulated deviations are evaluated at repetition completion boundaries using cumulative user movement data of the identified exercise to detect errors in the exercise movement of the user 304, wherein the DFTA 303 identifies incorrect form arising from improper execution or limited range of motion due to physical constraints or body structure anomalies; provide a real-time corrective media feedback to the user 304, by the DFTA 303, based on the detected errors to assist the user 304 in maintaining proper form; classify the exercise performance of the user 304 based on the tracked movements and the detected errors, wherein the DFTA 303 grades user skill levels and adapts feedback based on repetition accuracy, consistency, and speed of the user 304; and estimate energy expenditure and dynamically maintain a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics, wherein the DFTA 303 autonomously regulates energy targets in response to variations in exercise difficulty and user progression to achieve a personalized fitness outcome.

[0197] The computer program instructions described herein implement the processes of the various embodiments and may perform additional steps necessary to execute real-time exercise tutoring. When executed by the processor(s) 301, the instructions cause the processor(s) 301 to perform the steps of the method for real-time exercise tutoring, as described with respect to FIGS. 1A-1B. In some embodiments, a single program or a combination of programs implements one or more steps of the method. The processor(s) 301 retrieve and execute the relevant instructions from the memory unit 302.

[0198] A module, or an engine, or a unit, as used herein, refers to any combination of hardware, software, and / or firmware. As an example, a module, or an engine, or a unit includes hardware such as a microcontroller, associated with a non-transitory, computer-readable storage medium to store computer program codes adapted to be executed by the microcontroller. Therefore, references to a module, or an engine, or a unit, in an embodiment, refer to the hardware that is specifically configured to recognize and / or execute the computer program codes to be held on a non-transitory, computer-readable storage medium. In an embodiment, the computer program codes comprising computer readable and executable instructions are implemented in any programming language, for example, C, C++, C#, Java®, JavaScript®, Fortran, Ruby, Perl®, Python®, Visual Basic®, hypertext pre-processor (PHP), Microsoft®.NET, Objective-C®, the Swift™ programming language of Apple Inc., etc. In another embodiment, other object-oriented, functional, scripting, and / or logical programming languages are also used. In an embodiment, the computer program codes or software programs are stored on or in one or more mediums as object code. In another embodiment, the term “module” or “engine” or “unit” refers to the combination of the microcontroller and the non-transitory, computer-readable storage medium. Often module or engine or unit boundaries that are illustrated as separate commonly vary and potentially overlap. For example, a module or an engine or a unit may share hardware, software, firmware, or a combination thereof, while potentially retaining some independent hardware, software, or firmware. In various embodiments, a module or an engine or a unit includes any suitable logic.

[0199] It is apparent in different embodiments that the various methods, algorithms, and computer-readable programs disclosed herein are implemented on non-transitory, computer-readable storage media appropriately programmed for computing devices. The non-transitory, computer-readable storage media participate in providing data, for example, instructions that are read by a computer, a processor, or a similar device. In different embodiments, the “non-transitory, computer-readable storage media” also refer to a single medium or multiple media, for example, a centralized database, a distributed database, and / or associated caches and servers that store one or more sets of instructions that are read by a computer, a processor, or a similar device. The “non-transitory, computer-readable storage media” also refer to any medium capable of storing or encoding a set of instructions for execution by a computer, a processor, or a similar device and that causes a computer, a processor, or a similar device to perform any one or more of the steps of the method disclosed herein. In an embodiment, the computer programs that implement the method and algorithms disclosed herein are stored and transmitted using a variety of media, for example, the computer-readable media in various manners. In an embodiment, hard-wired circuitry or custom hardware is used in place of, or in combination with, software instructions for implementing the processes of various embodiments. Therefore, the embodiments are not limited to any specific combination of hardware and software. Various aspects of the embodiments disclosed herein are implemented in a non-programmed environment comprising documents created, for example, in a hypertext markup language (HTML), an extensible markup language (XML), or other format that render aspects of a graphical user interface (GUI) or perform other functions, when viewed in a visual area or a window of a browser program. Various aspects of the embodiments disclosed herein are implemented as programmed elements, or non-programmed elements, or any suitable combination thereof.

[0200] Where databases are described such as the database 305 exemplarily illustrated in FIG. 3, it will be understood by one of ordinary skill in the art that (i) alternative database structures to those described may be employed, and (ii) other memory structures besides databases may be employed. Any illustrations or descriptions of any sample databases disclosed herein are illustrative arrangements for stored representations of information. In an embodiment, any number of other arrangements are employed besides those suggested by tables illustrated in the drawings or elsewhere. In another embodiment, despite any depiction of the databases as tables, other formats including relational databases, object-based models, and / or distributed databases are used to store and manipulate the data types disclosed herein. In an embodiment, object methods or behaviors of a database are used to implement various processes such as those disclosed herein. In another embodiment, the databases are, in a known manner, stored locally or remotely from a device that accesses data in such a database. In embodiments where there are multiple databases, the databases are integrated to communicate with each other for enabling simultaneous updates of data linked across the databases, when there are any updates to the data in one of the databases.

[0201] The embodiments disclosed herein are configured to operate in a network environment comprising one or more computers that are in communication with one or more devices via one or more networks. In an embodiment, the computers communicate with the devices directly or indirectly, via a wired medium or a wireless medium such as the Internet, satellite internet, a local area network (LAN), a wide area network (WAN) or the Ethernet, or via any appropriate communications mediums or combination of communications mediums. Each of the devices comprises processors that are adapted to communicate with the computers. In an embodiment, each of the computers is equipped with a network communication device, for example, a network interface card, a modem, or other network connection device suitable for connecting to a network. Each of the computers and the devices executes an operating system. While the operating system may differ depending on the type of computer, the operating system provides the appropriate communications protocols to establish communication links with the network. Any number and type of machines may be in communication with the computers.

[0202] The embodiments disclosed herein are not limited to a particular computer system platform, processor, operating system, or network. One or more of the embodiments disclosed herein are distributed among one or more computer systems, for example, servers configured to provide one or more services to one or more client computers, or to perform a complete task in a distributed system. For example, one or more of embodiments disclosed herein are performed on a client-server system that comprises components distributed among one or more server systems that perform multiple functions according to various embodiments. These components comprise, for example, executable, intermediate, or interpreted code, which communicate over a network using a communication protocol. The embodiments disclosed herein are not limited to be executable on any particular system or group of systems, and are not limited to any particular distributed architecture, network, or communication protocol.

[0203] The foregoing examples and illustrative implementations of various embodiments have been provided merely for explanation and are in no way to be construed as limiting of the embodiments disclosed herein. While the embodiments have been described with reference to various illustrative implementations, drawings, and techniques, it is understood that the words, which have been used herein, are words of description and illustration, rather than words of limitation. Furthermore, although the embodiments have been described herein with reference to particular means, materials, techniques, and implementations, the embodiments herein are not intended to be limited to the particulars disclosed herein; rather, the embodiments herein extend to all functionally equivalent structures, methods, systems, and uses, such as are within the scope of the appended claims. It will be understood by those skilled in the art, having the benefit of the teachings of this specification, that the embodiments disclosed herein are capable of modifications and other embodiments may be effected and changes may be made thereto, without departing from the scope of the embodiments disclosed herein.

Claims

1. A method for real-time exercise tutoring, the method employing a dynamic fitness training assistant (DFTA), wherein the DFTA comprises one or more modules executable by at least one processor of a user device, and wherein the modules comprise computer program instructions for performing the method, comprising:prompting a user to activate a front camera of the user device to capture a live video of the user performing an exercise, and allowing the user to visualize the exercise performance on a graphical user interface of the user device as the user performs the exercise;analyzing the captured live video using a pose estimation model and a pose estimation algorithm of the DFTA to detect and extract key points of the user, wherein the DFTA analyses each frame of the captured live video;verifying orientation and visibility conditions through guided interaction by providing directional prompts to position the user for optimal key point detection, overlaying the optimal key points onto the analyzed frames of the live video of the user upon reaching a predefined detection confidence, and thereafter verifying the full-body visibility of the user by executing a key point-based visibility check based on the overlaid optimal key points, and, upon successful verification, the DFTA calculates joint angles of the detected and extracted key points to monitor posture and exercise movement of the user;creating a virtual workout box around the user based on an initial pose detected in the live video and maintaining the created virtual workout box as a persistent spatial constraint during performance of the exercise by anchoring the created virtual workout box to the initial user pose, without recalculating the created virtual workout box boundaries on a per-frame basis, wherein the created virtual workout box defines a region of interest for exercise tracking, wherein the DFTA creates the virtual workout box according to user-specific parameters and exercise-specific configuration data, and maintains the created virtual workout box unchanged until completion of each exercise;determining whether the live video of the user is within the created virtual workout box, monitoring and analyzing the exercise movement of the user using the detected and extracted key points and the calculated joint angles, wherein the DFTA pauses and resumes data analysis based on user position relative to the created virtual workout box and the front camera of the user device;tracking the exercise movement of the user using the detected and extracted key points and the calculated joint angles to identify type of exercise performed by the user and count number of repetitions, wherein the DFTA continuously analyzes motion data to ensure precise motion capture;identifying deviations from a predefined exercise motion pattern, wherein the deviations detected during an active movement phase are accumulated and used to generate corrective feedback, and the accumulated deviations are evaluated at repetition completion boundaries using cumulative user movement data of the identified exercise to detect errors in the exercise movement of the user, wherein the DFTA identifies incorrect form arising from improper execution or limited range of motion due to physical constraints or body structure anomalies;providing a real-time corrective media feedback to the user, by the DFTA, based on the detected errors to assist the user in maintaining proper form;classifying the exercise performance of the user based on the tracked movements and the detected errors, wherein the DFTA grades user skill levels and adapts feedback based on repetition accuracy, consistency, and speed of the user; andestimating energy expenditure and dynamically maintaining a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics, wherein the DFTA autonomously regulates energy targets in response to variations in exercise difficulty and user progression to achieve a personalized fitness outcome.

2. The method of claim 1, further comprising:preparing the DFTA for the real-time exercise tutoring by setting up all necessary components, comprising initializing variables, configuring the front camera of the user device, verifying permissions, and connecting to a cloud-based server for obtaining user-specific data.

3. The method of claim 2, wherein the user-specific data comprises personal information of the user, and wherein the personal information of the user comprises age, weight, height, injuries, and health and fitness level of the user.

4. The method of claim 1, wherein the DFTA is configured to capture the live video at a predetermined frame rate, based on confidence in estimating lightning conditions, to facilitate seamless tracking and enhancing the repetition accuracy of pose estimation.

5. The method of claim 1, wherein the real-time corrective media feedback comprises one or more of audio, video, text, and multi-media feedback.

6. The method of claim 1, wherein the DFTA identifies two states comprising rest state and active state while tracking the exercise movement of the user, wherein the rest state occurs when the user is in a starting position with muscles relaxed, and wherein the active state occurs when the user begins movement or is already performing the exercise.

7. The method of claim 6, wherein the DFTA detects a transition from the rest state to the active state when the value of the calculated joint angles change significantly.

8. The method of claim 7, wherein the DFTA counts the repetitions of the exercise being performed by tracking the exercise movement of the user from the rest state to the active state to back to the rest state.

9. The method of claim 8, wherein the DFTA classifies the exercise performance of the user into three categories comprising a novice, a skilled user, and a mastered user, based on tracked movements, repetition accuracy, consistency, and the detected errors.

10. The method of claim 9, wherein the DFTA tracks the exercise performance of the user over a predefined duration and modifies the user's future workout based on user's performance level, wherein the DFTA creates a dynamic artificial intelligence (AI)-driven workout plan that adapts based on the exercise performance of the user.

11. The method of claim 1, wherein, upon completion of each exercise, the DFTA generates an interactive post-exercise analysis interface configured to present performance insights and feedback indicators reflective of user activity, accuracy, and progression.

12. The method of claim 1, wherein the DFTA is configured to maintain calorie balance even if difficulty of the exercise is changed.

13. The method of claim 1, wherein the DFTA generates daily, weekly, monthly and long-term exercise summary reports.

14. The method of claim 1, wherein the DFTA is implemented as an on-device software installed and run on the user device.

15. A user device for real-time exercise tutoring, comprising:at least one processor;a memory unit for storing a dynamic fitness training assistant (DFTA) application, wherein the DFTA application comprises computer program instructions that, when executed by the at least one processor of the user device, causes the at least one processor to:prompt a user to activate a front camera of the user device to capture a live video of a user performing an exercise, and allow the user to visualize the exercise performance on a graphical user interface of the user device as the user performs the exercise;analyze the captured live video using a pose estimation model and a pose estimation algorithm of the DFTA to detect and extract key points of the user, wherein the DFTA analyses each frame of the captured live video;verify orientation and visibility conditions through guided interaction by providing directional prompts to position the user for optimal key point detection, overlay the optimal key points onto the analyzed frames of the live video of the user upon reaching a predefined detection confidence, and thereafter verify the full-body visibility of the user by executing a key point-based visibility check based on the overlaid optimal key points, and, upon successful verification, the DFTA calculates joint angles of the detected and extracted key points to monitor posture and exercise movement of the user;create a virtual workout box around the user based on an initial pose detected in the live video and maintaining the created virtual workout box as a persistent spatial constraint during performance of the exercise by anchoring the created virtual workout box to the initial user pose, without recalculating the created virtual workout box boundaries on a per-frame basis, wherein the created virtual workout box defines a region of interest for exercise tracking, wherein the DFTA creates the virtual workout box according to user-specific parameters and exercise-specific configuration data, and maintains the created virtual workout box unchanged until completion of each exercise;determine whether the live video of the user is within the created virtual workout box, monitor and analyze the exercise movement of the user using the detected and extracted key points and the calculated joint angles, wherein the DFTA pauses and resumes data analysis based on user position relative to the created virtual workout box and the front camera of the user device;track the exercise movement of the user using the detected and extracted key points and the calculated joint angles to identify type of exercise performed by the user and count number of repetitions, wherein the DFTA continuously analyzes motion data to ensure precise motion capture;identify deviations from a predefined exercise motion pattern, wherein the deviations detected during an active movement phase are accumulated and used to generate corrective feedback, and the accumulated deviations are evaluated at repetition completion boundaries using cumulative user movement data of the identified exercise to detect errors in the exercise movement of the user, wherein the DFTA identifies incorrect form arising from improper execution or limited range of motion due to physical constraints or body structure anomalies;provide a real-time corrective media feedback to the user based on the detected errors to assist the user in maintaining proper form;classify the exercise performance of the user based on the tracked movements and the detected errors, wherein the DFTA grades user skill levels and adapts feedback based on repetition accuracy, consistency, and speed of the user; andestimate energy expenditure and dynamically maintaining a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics, wherein the DFTA autonomously regulates energy targets in response to variations in exercise difficulty and user progression to achieve a personalized fitness outcome.

16. The user device of claim 15, further comprises:preparing the DFTA for the real-time exercise tutoring by setting up all necessary components, comprising initializing variables, configuring the front camera of the user device, verifying permissions, and connecting to a cloud-based server for obtaining user-specific data.

17. The user device of claim 15, wherein the DFTA classifies the exercise performance of the user into three categories comprising a novice, a skilled user, and a mastered user, based on tracked movements, repetition accuracy, consistency, and the detected errors, and wherein the DFTA tracks the exercise performance of the user over a predefined duration and modifies the user's future workout based on user's performance level, wherein the DFTA creates a dynamic artificial intelligence (AI)-driven workout plan that adapts based on the exercise performance of the user.

18. The user device of claim 15, wherein, upon completion of each exercise, the DFTA generates an interactive post-exercise analysis interface configured to present performance insights and feedback indicators reflective of user activity, accuracy, and progression.

19. A non-transitory, computer-readable storage medium having embodied thereon, computer program instructions executable by at least one processor of a user device for real-time exercise tutoring, said computer program instructions when executed by said at least one processor cause said at least one processor to:prompt a user to activate a front camera of the user device to capture a live video of a user performing an exercise, and allow the user to visualize the exercise performance on a graphical user interface of the user device as the user performs the exercise;analyze the captured live video using a pose estimation model and a pose estimation algorithm of the DFTA to detect and extract key points of the user, wherein the DFTA analyses each frame of the captured live video;verify orientation and visibility conditions through guided interaction, provide directional prompts to position the user for optimal key point detection, overlay the optimal key points onto the analyzed frames of the live video of the user upon reaching a predefined detection confidence, and thereafter verify the full-body visibility of the user by executing a key point-based visibility check based on the overlaid optimal key points, and, upon successful verification, the DFTA calculates joint angles of the detected and extracted key points to monitor posture and exercise movement of the user;create a virtual workout box around the user the user based on an initial pose detected in the live video and maintaining the created virtual workout box as a persistent spatial constraint during performance of the exercise by anchoring the created virtual workout box to the initial user pose, without recalculating the created virtual workout box boundaries on a per-frame basis, wherein the created virtual workout box defines a region of interest for exercise tracking, wherein the DFTA creates the virtual workout box according to user-specific parameters and exercise-specific configuration data, and maintains the created virtual workout box unchanged until completion of each exercise;determine whether the live video of the user is within the created virtual workout box, monitor and analyze the exercise movement of the user using the detected and extracted key points and the calculated joint angles, wherein the DFTA pauses and resumes data analysis based on user position relative to the created virtual workout box and the front camera of the user device;track the exercise movement of the user using the detected and extracted key points and the calculated joint angles to identify type of exercise performed by the user and count number of repetitions, wherein the DFTA continuously analyzes motion data to ensure precise motion capture;identify deviations from a predefined exercise motion pattern, wherein the deviations detected during an active movement phase are accumulated and used to generate corrective feedback, and the accumulated deviations are evaluated at repetition completion boundaries using cumulative user movement data of the identified exercise to detect errors in the exercise movement of the user, wherein the DFTA identifies incorrect form arising from improper execution or limited range of motion due to physical constraints or body structure anomalies;provide a real-time corrective media feedback to the user based on the detected errors to assist the user in maintaining proper form;classify the exercise performance of the user based on the tracked movements and the detected errors, wherein the DFTA grades user skill levels and adapts feedback based on repetition accuracy, consistency, and speed of the user; andestimate energy expenditure and dynamically maintaining a calorie balance based on user-specific metabolic characteristics and adaptive performance analytics, wherein the DFTA autonomously regulates energy targets in response to variations in exercise difficulty and user progression to achieve a personalized fitness outcome.

20. The non-transitory, computer-readable storage medium of claim 21, further comprising additional computer program instructions executable by the at least one processor of the user device, said additional computer program instructions when executed by said at least one processor cause said at least one processor to:prepare the DFTA for the real-time exercise tutoring by setting up all necessary components, comprising initializing variables, configuring the front camera of the user device, verifying permissions, and connecting to a cloud-based server for obtaining user-specific data.