Human Activity Recognition using Wearable Sensors | CNN Machine Learning Project | Google Colab
This video presents my Machine Learning project on Human Activity Recognition using wearable sensor data.
The goal of this project is to classify different human activities using accelerometer readings collected from wearable devices. A 1-Dimensional Convolutional Neural Network (1D CNN) model was developed and trained using sequential sensor data.
The dataset consists of six human activities:
• Cycling
• Push-up
• Run
• Squat
• Table Tennis
• Walk
Each sample contains accelerometer readings from five sensors, where each sensor measures movement along three axes (X, Y, Z). Therefore, each data point includes 15 features.
In this project, the following steps were performed:
• Data preprocessing and cleaning
• Sequence generation using overlapping windows
• Training a 1D CNN deep learning model
• Evaluating model performance using accuracy, confusion matrix, precision, recall, and F1-score
• Deploying the trained model using Hugging Face Spaces and Gradio
The trained model achieved approximately 89% classification accuracy on the test dataset.
Project Demo (Hugging Face Space):
https://huggingface.co/spaces/tharidux20/E...
Tools and Technologies Used:
• Python
• Google Colab
• TensorFlow / Keras
• Scikit-learn
• Hugging Face Spaces
• Gradio
This project demonstrates how deep learning techniques can be applied to wearable sensor data for activity recognition.