Fitness Tracker with Sensor Data is a machine learning-powered system designed to enhance workout tracking by analyzing motion data from sensors. It leverages advanced data processing, feature engineering, and predictive modeling to evaluate exercise form, classify workout types, and count repetitions. The project supports exercises like Squats, Bench Press, Deadlift, Overhead Press, and Barbell Row.

The system identifies and categorizes exercises based on sensor data, achieving high accuracy through machine learning models such as Random Forests, SVM, and Neural Networks.

Tracks exercise repetitions using advanced signal processing techniques, ensuring accurate real-time performance monitoring.

Provides feedback on form by analyzing motion and detecting improper postures, reducing the risk of injury.

Designed to work with wearable devices, enabling real-time tracking and insights during workouts.

• Cleansing: Standardizes raw accelerometer and gyroscope data.
• Outlier Removal: Employs methods like Interquartile Range (IQR) and Local Outlier Factor (LOF) to filter anomalies.

• Signal Transformation: Applies low-pass filtering and Fourier Transformations for noise reduction and frequency analysis.
• Dimensionality Reduction: Utilizes Principal Component Analysis (PCA) to enhance computational efficiency.
• Temporal Abstraction: Captures trends across time intervals for better context.

• Trains machine learning algorithms to classify exercises and count repetitions accurately.
• Optimizes model performance using hyperparameter tuning and cross-validation techniques.

1. Clone the Repository

   git clone https://github.com/Aishjahan/Fitness-Tracker-With-Sensor-Data.git
   cd Fitness-Tracker-With-Sensor-Data

2. Install dependencies from requirements.txt or environment.yml.

   pip install -r requirements.txt

3. Prepare the Dataset Place raw sensor data in the designated directory and preprocess it by running the data processing scripts.

Run the make_dataset.py script to preprocess the raw sensor data:

python src/data/make_dataset.py

Run build_features.py to create advanced features:

python src/features/build_features.py

Use train_model.py to train and optimize classification models:

python src/models/train_model.py

Run count_repetitions.py to apply filters and count repetitions:

python src/features/count_repetitions.py

Generate workout and feature visualizations with visualize.py:

python src/visualization/visualize.py

The classification model achieves 98% accuracy, offering highly reliable detection of exercise types and ensuring precise tracking of workout patterns for enhanced fitness insights.

TrackFit AI leverages machine learning and sensor data to revolutionize personal fitness tracking by providing accurate exercise classification, posture correction, and repetition counting. With 98% accuracy in exercise recognition and advanced visualization tools, this project bridges the gap between technology and fitness. By promoting safe and effective workouts, TrackFit AI ensures users achieve their fitness goals with confidence. Future enhancements, such as integration with additional wearables and personalized feedback systems, will further elevate its utility and impact on the fitness community.