MultiSensing®
Human Intelligence and Flexibility for Your System.
MultiSensing® is Neonode's proprietary technology for computer vision solutions.
Utilizing the latest advances in machine learning in combination with our in-house synthetic data factory, Neonode’s camera-based MultiSensing solution can interpret and understand a wide range of objects, situations and contexts from real-time image data.
Using purely synthetic data to train our neural networks, we are in absolute control of both the images shown to the network and the labels needed for training. Transcending manually annotated real-world data, MultiSensing is able to go beyond the limits of traditional neural networks in accuracy, scope, and efficiency.
USPs - MultiSensing
Discern the Relevant From Immaterial
Program MultiSensing to recognize anything or anyone. Teach it to understand the things it must see and what it should ignore.
Rapid Iteration of Data and Networks
Neonode's in-house render-farm can produce massive datasets in just hours, enabling rapid iterations on both data and networks.
Camera Position and Lens Agnostic
MultiSensing accepts camera inputs from RGB, IR, TOF, pinhole, wide-angle or fisheyes. It runs effectively in all standard camera positions.
Incorporate Proprietary CAD Models
Use your own CAD designs to fine tune existing MultiSensing solutions or create brand new ones catering to specific requirements and needs.
Neonode's MultiSensing Software
MultiSensing’s core software is built according to automotive ASPICE software and project practices, leading to robust yet highly-flexible software. On this foundation, expansions and customer-specific addons are easy to introduce and maintain.
How Neonode Leverages Synthetic Data to Create Best in Class DMS Functionality
By generating synthetic data and controlling the whole machine learning pipeline, Neonode's MultiSensing software platform is built to deliver best-in-class performance for Driver and In-Cabin Monitoring.
Learn how our use of synthetic data allows us to generate features that would not be possible using real world inputs.