SensAKitchen: Contextualizing Kitchen Activities with IoT-Enabled Event Identification

This study demonstrates how environmental sensors are effectively used to find the context of different types of cooking activities. This work was conducted in a household kitchen with proper ventilation and using a sensor array comprised of PM
2.5
, PM
10
, CO
2
, Temperature and Humidity sensors to investigate different cooking methods, such as boiling, frying. Several machine learning models were used, with and without hyperparameter tuning, for the evaluation of experimental results. The overall accuracy obtained is around 90% and above 95%, before and after using hyper-parameter tuning, respectively. These results show the efficacy of environmental sensors coupled with machine learning techniques can accurately identify and contextualise various cooking activities within a home kitchen environment. Such insights promise to enhance the remote monitoring of cooking practices in domestic settings for smart kitchen environments.