The monitoring of indoor air quality (IAQ) is essential to prevent its potential effect on our health. One of the main sources of pollutants is the daily occupant activities, such as the use of cleaning products and cooking. To avoid that the levels of indoor pollution induced by these sources get harmful, and to aware people of their occurrence and impact, indoor air pollution events should be identified through the use of gas sensors. This work proposes a method of detection of indoor pollution activities through the estimation of the intrinsic dimensionality on windows of observation of metal-oxide gas (MOX) sensors. This approach was tested on a dataset extracted from 2 months of experiments in which typical household activities were performed in a 17m² room using 21 unique models of MOX sensors. Results show that this method is capable of detecting activities in the environment but more understanding is needed to fully characterize it