Analysing Cooking Behaviour in Home Settings: Towards Health Monitoring

Kristina Yordanova (1, 2) , Stefan Lüdtke (1) , Samuel Whitehouse (2) , Frank Krüger (1) , Adeline Paiement (2, 3) , Majid Mirmehdi (4) , Ian Craddock (2) , Thomas Kirste (1)

1 Universität Rostock
2 University of Bristol [Bristol]
3 DYNI - DYNamiques de l’Information
4 Computer Science Department [Bristol]

Wellbeing is often affected by health-related conditions. One type of such conditions are 1 nutrition-related health conditions, which can significantly decrease the quality of life. We envision a 2 system that monitors the kitchen activities of patients and that based on the detected eating behaviour 3 could provide clinicians with indicators for improving a patient's health. To be successful, such 4 system has to reason about the person's actions and goals. To address this problem, we introduce a 5 symbolic behaviour recognition approach, called Computational Causal Behaviour Models (CCBM). 6 CCBM combines symbolic representation of person's behaviour with probabilistic inference to reason 7 about one's actions, the type of meal being prepared, and its potential health impact. To evaluate the 8 approach, we use a cooking dataset of unscripted kitchen activities, which contains data from various 9 sensors in a real kitchen. The results show that the approach is able to reason about the person's 10 cooking actions. It is also able to recognise the goal in terms of type of prepared meal and whether it 11 is healthy. Furthermore, we compare CCBM to state of the art approaches such as Hidden Markov 12 Models (HMM) and decision trees (DT). The results show that our approach performs comparable to 13 the HMM and DT when used for activity recognition. It outperforms the HMM for goal recognition 14 of the type of meal with median accuracy of 1 compared to median accuracy of 0.12 when applying 15 the HMM. Our approach also outperforms the HMM for recognising whether a meal is healthy with 16 a median accuracy of 1 compared to median accuracy of 0.5 with the HMM. 17

Mots clés: plan recognition; activity recognition; behaviour monitoring; symbolic models; probabilistic models; goal recognition; sensor-based reasoning

Domaines: Intelligence artificielle [cs.AI]; Apprentissage [cs.LG]; Traitement du signal et de l'image [eess.SP]

Kristina Yordanova, Stefan Lüdtke, Samuel Whitehouse, Frank Krüger, Adeline Paiement, et al.. Analysing Cooking Behaviour in Home Settings: Towards Health Monitoring. Sensors, In press. ⟨hal-02003387⟩