Automated recognition of hair touch and scalp scratching through novel sensing modalities, sensor fusion and machine learning approaches

Computational behaviour analytics aims to recognise everyday human activities and the context in which they occur to obtain quantifiable information about user's daily routines, habits and reactions to potential interventions. Within this context, the automated detection of how people touch parts of their body, specifically hair and scalp have significant applications in Unilever.
This project aims to build and validate wearable technology to measure people's reaction to scalp itch and most importantly, asses the effect of personal care products in this context.
The aim is to develop an approach to automatically recognise the action of hair and scalp touch/scratching
(a) Defining the relevant types of interactions and the main qualitative and quantitative variables which serve as measure for itch.
b) Identify existing and novel sensor modalities and activity recognition approaches which are optimal for the detection challenge.
(c) Investigate the trade-offs between recognition performance and sets of modalities and number of devices, as relevant for cost and user comfort.
(d) Evaluate the suitability of each combination of modalities in a realistic context.
Work Plan
Part 1: Survey of scalp touch interaction and