Can YOU predict activity from sensor data?

A challenge has been launched for people to predict posture and movement within a smart home from video, accelerometer and environmental sensor data. The competition is part of the University of Bristol’s SPHERE project, which is using a unique platform of sensors to quantify health-related behaviours over long periods and also to help older people in the future to live safely at home while maintaining their privacy and independence.

The SPHERE (Sensor Platform for HEalthcare in a Residential Environment) Challenge: Activity recognition with multimodal sensor data is being run in collaboration with ECML-PKDD 2016 and DataDriven. Deadline for entries to the competition is 31 July 2016 and winners will be announced at the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery (ECML-PKDD) 2016 conference in September.

Volunteers undertook a series of ‘activities of daily life’ in SPHERE’s smart house in Bristol, which is equipped with a purpose-built network of advanced 3D cameras, wearables and ambient sensors.

The task for the SPHERE challenge is to predict posture and ambulation labels given the sensor data of 20 specifically chosen activities, such as: walk, sit, stand-to-bend, ascend stairs, descend stairs, from the volunteers.

Professor Ian Craddock, Director of SPHERE-IRC based in the Faculty of Engineering, said:

“The fastest rising categories of health costs in the UK are associated with long-term health conditions such as diabetes, dementia and depression. Advanced sensing and artificial intelligence technologies are capable of revealing long-term behavioural patterns that are important in understanding the progression and management of illness, especially in ageing populations.

“The competition is setting people a challenge to help push forward the state-of-the-art by predicting actual activity from sensor data. We’re extremely grateful to the AARP Foundation for supporting literally hundreds of talented researchers around the world to work with the first of SPHERE’s unique datasets.”

A prize fund of $10,000, sponsored by the AARP Foundation USA that works to ensure low-income older adults have nutritious food, safe and affordable housing, a steady income, and strong and sustaining social bonds, will be awarded as follows:

• $5,000 being awarded to the winner
• $3,000 to the runner up
• $2,000 to the second runner up

There is also an additional prize fund for the three best-scoring entrants that attend ECML-PKDD 2016, with €1,000 being awarded to the highest scoring attendee; €600 to the second highest scoring attendee; and €400 to the third highest scoring attendee. These prizes will be awarded at the conference.

Paper

‘The SPHERE challenge activity recognition with multimodal sensor data’ by Niall Twomey, Tom Diethe, Meelis Kull, Hao Song, Massimo Camplani, Sion Hannuna, Xenofon Fafoutis, Ni Zhu, Pete Woznowski, Peter Flach, and Ian Craddock published in arXiv.