- Jon Colosimo
Use Your Camera As A Sensor
One Artificial Intelligence application that I've been coming across is using image-based data to detect an event related to an industry's typical need of observance - also know as Machine Vision. This technology eliminates the need of human monitoring (that guy with a clipboard that counts how many cars go left and right out of the parking lot.
With data being harnessed from live and archived video streams, sensing anything with a camera is a possibility these days. Dagostino Electronic Services' Security Department offers state of the art camera and video management solutions for security purposes, and the technologies they use are constantly evolving allowing users to do more with the data than just archive. Some examples of solutions created and provided to customers are:
Space utilization (people counting)
Event recognition (someone hopping over a fence)
Temperature sensing (we all remember having our temperatures taken EVERYWHERE when COVID was at it's peak)
License Plate Recognition, vehicle make, and vehicle color
Machine Vision in Action
How does Machine Vision interpret camera data? Let's circle around the use case of autonomous vehicles.
An autonomous and self-driving car, such as a Tesla, is covered with sensors all around the exterior to capture the vehicle's surroundings. Some of those sensors are typical and single purpose, but the real magic is in the camera technology used as sensors. Per Tesla's autopilot information, "Each new Tesla vehicle is equipped with eight external cameras and powerful vision processing to provide an additional level of security". When they say security, they really mean safety!
Think about driving in the winter and your reaction if you see a potential icy or slippery part of the road. Human vision has the capability to see and process information at a rate of about 30-60 frames per second (fps). By leveraging computers and cameras, the camera can capture at frames rates up to 225 fps, being able to recognize, process, and react almost 4 times faster that the human can when time is extremely important!
Check out how Artificial Intelligence Researcher Lex Fridman demonstrated Machine Vision technology inside and outside of an autonomous vehicle: