Introducing the End-to-end Computer Vision Solution for the Built and Retail Environments
Physical space is something that many of us take for granted, but the events of the last year have redefined how we treat it. As companies look for better ways to use and manage their physical space — and conduct business within it — they should look no further than computer vision, which can offer an unprecedented level of insight about the nuances of what happens within their stores, office buildings and other facilities. But building a CV solution takes a tremendous amount of time and technical know-how. It also requires a multiplicity of decisions every step of the way, from which framework and machine learning model to use, to how many sensors to deploy, to where to put them. Taken together, each of these decisions adds up to a development process that can last weeks, or even months. (For more on that process, check out my last post.)
That’s why at Nomad Go we built an end-to-end CV solution that delivers insights out of the box and removes the complexity that often keeps CV deployments from succeeding in the real world. You establish the use case (e.g., understanding building occupancy, end-cap dwell time, etc.), and Nomad Go will determine the best CV model and the best way to capture the data, as well as provide comprehensive analytics through a dashboard that can be used any way you need (e.g., reports, controlling building system, etc.) to get the actionable insights to improve your business.
Supporting multiple CV frameworks and machine learning models
Success with CV requires computer vision models to detect exactly what you want them to see and avoid detecting the things you don’t want. Our system supports our own machine learning models as well as a variety of third-party frameworks and models from Apple, NVIDA and others, so we can support a wide range of scenarios.
Achieving on-device data processing at the network’s edge
Unlike any other CV solution on the market, Nomad Go uses smart devices and smart cameras that run on the network’s edge. This gives you a cost-effective solution with the raw, on-board computing power to process visual data on the device itself. And that’s no small thing.
Processing data in the cloud comes with additional costs that are simply unavoidable: the cost of building a big enough pipe to upload your data, the continued costs of analyzing and storing that data, and possibly the cost of installing and managing 2 or 3 servers in each of your stores or buildings.
Processing at the edge virtually eliminates all of that. The result? Fewer network resources to manage, no privacy concerns and data analytics costs that are a fraction of what you’d pay in the cloud.
Organizing data into actionable metrics
Once operational, Nomad Go’s proprietary machine learning algorithm analyzes the data to identify any of the objects it has been trained to identify. Then the algorithm time-stamps and chronologically aggregates the data so you can identify patterns and trends as they emerge. No need to clean, organize or filter it. Instead, you’ll get actionable, pre-built metrics covering most any business scenario you can think of, straight out of the box.
And with Nomad Go’s extensible platform you have the flexibility to consume these insights in several ways:
• Monitor data and identify trends on the Nomad Go dashboard (via computer or smartphone app, or viewable on a monitor).
• Integrate with existing dashboards and displays using our API.
• Use the Message Queuing Telemetry Transport (MQTT) protocol to send data via low-bandwidth network connections to Internet of Things solutions, such as an HVAC control system.
For example, by connecting Nomad Go to an HVAC solution, you can adjust heating, cooling and ventilation based on occupant density instead of CO2 sensors, which are far less accurate and have a slow response time.
It’s easy to become overwhelmed at the thought of building your own CV solution. With an end-to-end solution from Nomad Go, you can have an operational CV solution in place in just a few hours, ready to collect data from your spaces and well on its way to giving you deep insights about how to improve your business.
In my next post I’ll give you a real-world example of what that looks like.