Computer Model Could Give Smart Cars "Human" Vision

A computer model developed by MIT neuroscientists that mimics the human vision system can accurately detect and classify cars, motorcycles, and other objects in a busy street scene. Among its many possible applications: smart sensors that can alert drivers to pedestrians and other dangers — not to mention robotic cars and other intelligent transport applications.

The model comprises 10 million computational units that behave like the clusters of neurons layered in the visual cortex. Images are fed into a learning algorithm, which extracts their common features. With each successive layer, more complex features and relationships are extracted; among other things, this enables the system to recognize the same object at different angles.

Neuroscientist and team member Thomas Serre spoke with MIT Technology Review:

"It was a surprise to us when we applied this model to real-world visual tasks and it competed well with the best systems," says Serre. Indeed, in some tests their model successfully recognized objects more than 95 percent of the time, on average. The more images the system is trained on, the more accurately it performs. [...]

While the current model analyzes only still images, the team is working on a parallel system that can handle video as well.

[Source: MIT Technology Review]