A monitoring system includes a first sensor and a second sensor, at least one of which is a camera. A computer system coupled to the first and second sensors includes a memory and stored object types and associations between objects. The computer system is configured to infer a group association between a first and second object by retrieving an input from at least one of the sensors, inferring an interaction at a monitored location, infer possession of the second object by the first object, and terminate the inferred group association based on a subsequent inference.