Michael Pehel

Jun 21, 2017 12:23:00 PM

Aerialtronics Pensar - Copy.jpg

This Monday Aerialtronics released Pensar, a twin-sensor computer vision platform for drones. Pensar combines a 30x zoom Sony daylight camera and a FLIR Boson thermal IR camera sensor data and parses it via a deep learning algorithm. The platform rests on the NVIDIA Jetson module for real-time processing and feedback.

With its real-time video processing, it can track objects for annotation, such as deviations in points of interest or blurring out people or signage for privacy reasons. Its computer vision algorithms are robust enough to read the characters on a license plate and virtually any other alphanumeric signage at a distance while in flight. Overall, the system decreases the time needed for post-processing and allows for in the moment emergency detection like gas leaks, or perimeter breeches when used for security surveillance.

License_Plate_Pensar - Copy.jpg

The platform runs on Ubuntu Linux, which makes it open-source and compatible with most major machine learning libraries. Its highly hackable nature makes it easy to customize for the specific needs of each operation, and adaptable enough to fit into the currently established workflows of any industry. Aeraltronics will also offer hand-coded solutions upon end-user request, a handy addition for companies without a developer on staff.

 

Related Search Term(s): Artificial Intelligence, Releases, Control Systems

Thoughts? Leave a comment: