Neovision2
Abstract
Biological vision systems have the exquisite ability to recognize, categorize, and learn new objects in fractions of a second. While animals and humans accomplish this seemingly effortlessly and constantly, computational vision systems have, to date, been unable to replicate this feat of biology. The Neovision2 program pursued an integrated approach to developing an advanced object recognition capability based on the visual pathways in the mammalian brain. Specifically, this program developed a cognitive sensor technology with limited size, weight, and power that transforms data from an imaging sensor suite into communicable knowledge for mobile, autonomous surveillance systems. The program demonstrated an improvement of four orders of magnitude in energy efficiency compared to state-of-the-art algorithms. To achieve the vision, the program utilized advanced device design, signal processing and mathematical techniques across multiple brain regions to create an electronic neuro-biological (neuromorphic) vision system.
Document Details
- Document Type
- Accomplishment
- Publication Date
- Oct 01, 2014
- Source ID
- 6869336a61e126499f17eb65ba0126ac