Download
LSTM-Human-Activity-Recognition is a machine learning project that demonstrates how recurrent neural networks can be used to recognize human activities from sensor data. The repository implements a deep learning model based on Long Short-Term Memory (LSTM) networks to classify physical activities using time-series data collected from wearable sensors. The project uses the well-known Human Activity Recognition dataset derived from smartphone accelerometer and gyroscope signals. Through the use of sequential neural network architectures, the system learns patterns in motion data that correspond to activities such as walking, sitting, standing, or climbing stairs. The repository includes data preprocessing scripts, neural network architecture definitions, and training pipelines that allow researchers to reproduce and modify the experiments. It serves as an educational example of how deep learning models can process temporal sensor signals for pattern recognition tasks.
Features
- Implementation of LSTM-based neural networks for time-series classification
- Human activity recognition using wearable sensor datasets
- Data preprocessing and feature extraction for accelerometer and gyroscope signals
- Training pipeline for sequential deep learning models
- Evaluation metrics for activity classification performance
- Example implementation demonstrating recurrent neural networks for sensor data
Project Samples
