Smart fitness suit for user assistance based on Machine Learning
Fitness and sport have become an important topic in wearable and pervasive computing. Regular exercises are known to be advantageous for one's health and subjective well-being, and is associated with improved physical fitness, lower mental stress, and anxiety levels. However, during high intensity a
2025-06-28 16:35:21 - Adil Khan
Smart fitness suit for user assistance based on Machine Learning
Project Area of Specialization Artificial IntelligenceProject SummaryFitness and sport have become an important topic in wearable and pervasive computing. Regular exercises are known to be advantageous for one's health and subjective well-being, and is associated with improved physical fitness, lower mental stress, and anxiety levels. However, during high intensity and forceful exercise sessions, there is a high probability that physical health is severely affected, jarring motions and improper posture during workouts can lead to temporary or permanent disability. Due to technological advances, activity recognition based on wearable sensors has attracted a large number of studies. Work is done in the field of smart gyms but none of the existing projects tackle serious problems as targeted by our system. “Athos Core” is considered to be the forefront of smart clothing even it monitors your muscle activity, heart rate and breathing rate but doesn’t make any sort of predictions to guide user during workouts. Our solution will obstruct this problem by predicting correct posture, monitoring muscle engagement and generating alerts on muscle fatigue. Smart fitness suit will assign a category of activity to the signal provided by wearable sensors such as accelerometer, gyroscope, and electromyography (EMG). Wearable sensors embedded with smart fitness suit will send raw data to the microcontroller in our case Arduino, the data will be classified and trained once based on machine learning algorithms. With the help of the fire base, android app will be connected to sensors and will guide user according to the context. It will keep on making alerts/warnings in case of muscle fatigue and will suggest a break. The purpose of monitoring muscle engagement is to let user know if he/she is engaging both arms equally to avoid uneven muscle growth. After making several predictions suit will guide the users who are not able to reach the right posture not to exercise i.e. people with incorrect alignment of backbone by birth. Last but not the least correct posture would avoid many serious injuries during workouts.
Project Objectives- To acquire datasets using a variety of sensors.
- To predict the most accurate posture.
- To develop an android application to guide the user and give alerts or warnings.
- To displace the need of fitness trainer
Accelerometer, gyroscope, and electromyography (EMG) sensors will be embedded in upper part of suit, as our focus is to target 6-7 back exercises for our project. A few voluntarily participants would be monitored one week for each exercise. Raw data would be collected through wearable sensors attached to their bodies. Features will be extracted from data to make it usable in algorithmic form. The next step would be the classification of data where machine learning algorithms will be used and a validation dataset would be generated that will be used for test data later on. Android applications will be developed. It will receive input as sensor values through fire base real time database and make prediction based on machine learning. Android text to speech feature will also be used for user guidance.
Benefits of the Project- It will help to exercise like a pro without the need of personal trainer.
- To let the women, exercise at home during the days they can’t get gym services.
- Correct form during workouts will save the user from injuries.
- Helping introverts to workout at home or at gym confidently.
Above figure depicts the overall project flow and the components used in each stage. To be more precise, with the help of wearable sensors (accelerometer, gyroscope, electromyography, microcontroller) embedded in suit, we will get real time sensor data. For training purpose, the datasets will be generated on things peak. Paid services are used to reduce latency up to 1 second. The datasets will be used for training a Machine Learning model that will help in recommendations. After training model will be deployed on AI Platform. At testing stage, data will first be transferred to fire base real time database and them synchronized in android app with a delay of milliseconds. Android app will request a prediction using cloud functions that connects to AI Platform where model is deployed. The model will predict the most accurate angel, recommendations will be made based on these predictions and thresholds. Android text to speech feature will be used for recommendation purpose. Finally, the programming languages used in the overall project are C++, Java, XML, Typescript and Python.
Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther Industries Education , Manufacturing , Others , Health Core Technology Internet of Things (IoT)Other Technologies Artificial Intelligence(AI), Cloud Infrastructure, OthersSustainable Development Goals Good Health and Well-Being for People, Industry, Innovation and Infrastructure, Partnerships to achieve the GoalRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 78900 | |||
| Arduino UNO (with power cable and casing) | Equipment | 3 | 900 | 2700 |
| Node MCU | Equipment | 2 | 450 | 900 |
| MyoWare Muscle Sensor (SEN-13723) | Equipment | 3 | 8500 | 25500 |
| 6DOF accelerometer/groscope sensor (MPU-6500) | Equipment | 4 | 750 | 3000 |
| EMG Electrodes | Equipment | 24 | 300 | 7200 |
| EB Muscle Sensor V1.2 | Equipment | 2 | 7000 | 14000 |
| ECG Electrodes | Equipment | 10 | 60 | 600 |
| Fitness Jacket with proper partition | Equipment | 2 | 7500 | 15000 |
| Cloud Services | Miscellaneous | 1 | 10000 | 10000 |