Smart Sensing and Tracking with Video and Mote Sensor Collaboration

With growing concerns about the security of our borders and vulnerable entry points video imagery has emerged as a high potential surveillance data source for generating real time situational information. This potential comes with its challenges due to the large volumes of data need to be processed, lack of smart sensors to automatically detect and understand video content, bandwidth limitations to distribute the video to the hands of decision makers to provide actionable intelligence for first responders. Our proposal addresses each of these challenges, including end-to-end system integration with specific attention to deployment and life cycle costs with Smart Sensing and Video Understanding in a sensor web platform: 1. We embed the video content analysis processing into the video sensor with our Smart Video Node to offload processing of video to the sensor edge; 2. We use inexpensive ad hoc netted sensor motes to cue video sensors to collect and process data only when a mote field is triggered based on commanded rules; 3. We use a sensorweb platform to quickly discover sensors that are available, seamlessly integrate and collect data from disparate sensors, perform system leverage, remotely task sensors and receive alerts from multiple sensors.