Recognizing human activity using multiple wearable accelerometer sensors

Human Activity Recognition, is the problem of predicting what kind of activity a person is performing based on a signals detected by smartphone sensors on their waist.

Two types of sensors present in smartphones are:

1. Accelerometer

2. Gyroscope

Accelerometer measures acceleration and Gyroscope measures angular velocity.

How the data was prepared?

1. 30 volunteers referred to as subjects performed the experiment for data collection wearing smartphones sensors on their waist.
2. The two smartphone sensors captured the 3 axial linear acceleration as well as the 3 axial angular velocity of the subject.
3. The sensor signals were sampled in fixed-width sliding windows of 2.56 sec and 50% overlap (128 readings/window).
4. The data were recorded at the constant frequency of 50Hz (50 data points were recorded each second )

Quick Overview of the Dataset

• Data is downloaded from following source: https://archive.ics.uci.edu/ml/datasets/human+activity+recognition+using+smartphones
• Feature names are present in UCI_HAR_dataset/features.txt
• Train Data
  • UCI_HAR_dataset/train/X_train.txt
  • UCI_HAR_dataset/train/subject_train.txt
  • UCI_HAR_dataset/train/y_train.txt
• Test Data
  • UCI_HAR_dataset/test/X_test.txt
  • UCI_HAR_dataset/test/subject_test.txt
  • UCI_HAR_dataset/test/y_test.txt

Problem Statement

Predict one of the following six activities that a Smartphone user is performing at that 2.56 Seconds time window by using either 561 feature data or raw features of 128 reading.

• Walking
• Walking Upstairs
• Walking Downstairs
• Sitting
• Standing
• Laying

LSTM architecture that will solve HAR problem

Project Contributors

• Kumar Shivam Ranjan
• Neha Kumari
• Madhav Bansal