Since 1964, more than 20 million Americans have died as a result of smoking. Being able to reflect on smoking habits allows smokers to quit more easily. By using a smartwatch features can be extracted as a tool to detect smoking gestures. Smoking gesture detection used machine learning to detect patterns from various features including max, median, mean, and variance speed, and net, median, and max roll velocity. The machine learning algorithm then classified which gesture was being performed based on training data of 30 people performing 4 distinct gestures 30 times each. The WEKA data mining library tested multiple classification algorithms and the most accurate was Random Forest. Successful detection of these smoking gestures can help smokers quit by providing real time interjections.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Ireland, Steven Patrick
• Romeo, Anthony A

Publisher

• Worcester Polytechnic Institute

Identifier

• E-project-042816-143740

Advisor

• Agu, Emmanuel O.

Year

• 2016

Date created

• 2016-04-28

Resource type

• Major Qualifying Project

Major

• Computer Science

Rights statement

• In Copyright