INVISION logo_White

We enable edge devices to interpret the world around them by recognizing objects, tracking their movements, and identifying anomalies.  Our AI software is so efficient that it runs on mainstream processors, making deployment to hundreds, or hundreds of thousands, of devices technically and financially viable today.

AI Accelerated

efficient-network@2x

We have taken a different approach, developing our own machine learning technology from the ground up.  Massively reducing both memory and computational footprint, while delivering accuracy equivalent to the best cloud systems.  We call this AI Software Acceleration, and it is the perfect complement for the latest generation of efficient low-power hardware.  Together we deliver a combination of performance and value that is unmatched.  Like any credible claim we have the data and competitive wins to back it up.  Contact us to find out more.

Modular Functions

We have developed a series of AI software modules that perform discrete functions, so that our customers can pick and choose the functions that they need for each specific device and application.  We support our customers’ diverse requirements from a single code base, so that ever customer benefits from ongoing performance improvements and new features.

Detection

Detection

Classification

Classification

Facial Recognition

Facial Recognition

Object Tracking

Object Tracking

Analytics

Analytics

Anomaly Detection

Anomaly Detection

Simplified Management

cloud-management@2x

Our cloud services make managing hundreds, or thousands, of edge devices easier.  We have partnered with industry leaders such as Microsoft and NXP Semiconductor to ensure security: from trusted edge hardware to secure cloud services.  Our networks are trained on the cloud, and that learning is distilled into compact firmware than can be manually or automatically pushed to edge devices.  As those edge devices detect, classify and track objects of interest they send lightweight metadata describing the event to the cloud.  Sophisticated analytics can be run in real-time across thousands of sensors.  All this information is available to your applications, securely and in one place, via a robust API.

Getting Started

We make it easy to evaluate our technology within your application environment.  This lets you quickly separate fact from hype, which is essential in developing a solid implementation plan.

PILOT

Pilots are less than 90 days duration, run on one of our supported platforms and use whatever sensors you have in mind for your application. We deliver a quantitative result set and a fully functioning application for you to run your own tests.

PRODUCTION

We help select the best sensors and processors for your edge devices.  We build any custom classifiers or advanced functions that you need and optimize the software for the target hardware.  We support integration of your applications, testing and roll-out.

SUPPORT

We provide patches, performance updates and the ability to add new functionality to existing hardware.  We can retrain the networks as scenarios not included in the previous training data are discovered, continuously improving accuracy.