AI-based camera provider StradVision is developing animal detection systems for its deep-learning SVNet software.
In the USA, there are more than 1.5 million deer-related accidents causing around 200 deaths and more than US$1bn of property damage.
Animal detection would allow auto makers to proactively stop the vehicle or steer the vehicle around the animal if the computer detects an immediate danger.
The function can detect animals in difficult conditions such as rain, snow, fog or darkness, and in urban situations.
Junhwan Kim, CEO of StradVision, said, “Upon detection of an animal, SVNet will be able to assume control of the vehicle virtually at the speed of light, in scenarios where human reactions are not going to prevent the collision. We are truly excited about the possibilities for this technology in the USA, where car-deer collisions are a problem, and in areas of Asia where animals are a common road obstacle.”
Kim added, “We have also considered environmental conditions. Even in a road next to trees and branches and other conditions that could hide animals, SVNet maintains a high level of accuracy. Other perception solutions are often confused in these situations and accuracy goes down, and you don’t want that when there’s a deer about to pop out of the woods.”