The present invention provides the multimodality filtration surveillance comprising of a plurality of filtration stages executed at the backend server to confirm nature of anomaly in an event, the filtration stages comprising: a first filter of a video anomalies detection in the event for a specified time-place value, a second filter of a city soundscape adapted to provide a localized decibel maps of a city, a third filter of a geocoded social network adapted to semantically read and analyze data from one or more social media corresponding to the specified time-place value, and a fourth filter of an event triggered or proactive local participatory surveillance adapted to provide augmented information on the detected anomalies.