Next-Gen Computer Vision
Dramatic improvement in Visual AI
Our patent pending technology is the first to combine Computer Vision and Physics Simulation to understand, effectively track, and predict vehicle movement, allowing us to improve the performance of Visual AI solutions and extend their range of applications.
How it works?
Input – 2 image
Our neural network understands cars
and curbside in 3-dimensions and from any angle
A 3-dimensional, vector space replica of the space is created in physics simulation
The Physics Simulation/Engine is developed in Unity environment to take advantage of existing calculations that are available and extensible with the scripting language. Physics Simulation allows us to build the vector space, understand it in 3-dimensions, apply laws of physics, predict the movement of Neural Network outputs and therefore eliminate impossible and improve the accuracy of neural network detections.
Benefits: View the captured scene from any angle. Understand object movement using physics simulation. High-precision understanding of temporarily occluded objects. The objects in the scene are temporary stable.
Real-world coordinates & revolution in motion tracking
We are connecting vector space with real world coordinates to get the accurate position of objects detected by the neural network. This allows us to move away from pixel-tracking and introduce location tracking, eliminating the need for high computational power otherwise needed.
It works by analyzing images captured by cameras mounted on stationary, moving, or flying objects and in real-time.