When you first explore the capabilities of our Groundlight AI detectors, you'll quickly notice that they excel at answering binary questions. These are queries expecting a straightforward "Yes" or "No" response. However, the world around us rarely conforms to such black-and-white distinctions, particularly when analyzing images. In reality, many scenarios present challenges that defy a simple binary answer.

We're thrilled to announce a new repository that makes it incredibly easy for anyone to get started for free with Groundlight, a computer vision (CV) platform powered by natural language. This first steps guide is designed to walk you through setting up your first Groundlight detector to answer a simple question: "Is the door open?" Behind the scenes, Groundlight will automatically train and deploy a computer vision model that can answer this question in real time. With our escalation technology, you don't need to provide any labeled data - you get answers from your first image submission.

Groundlight has a Problem​

Here at the Groundlight office we have a bit of a problem - sometimes we leave dirty dishes in the office sink. They pile up, and as the pile grows it becomes more and more tempting to simply add to the pile instead of cleaning it up. It was clear that the Groundlight office needed a “grime guardian” to save us from our messy selves. One day, I realized that this was the perfect problem to solve using Groundlight’s computer vision SDK. I could focus on developing the complex embedded application logic while Groundlight handled the computer vision. My design provided me with an opportunity to test out a handful of interesting design patterns, including deployment on a Raspberry Pi, multi-camera and multi-detector usage, a microservice-like architecture achieved via multithreading, and complex state handling.

At Groundlight, we put careful thought into measuring the correctness of our machine learning detectors. In the simplest case, this means measuring detector accuracy. But our customers have vastly different performance needs since our platform allows them to train an ML model for nearly any Yes/No visual question-answering task. A single metric like accuracy is unlikely to provide adequate resolution for all such problems. Some customers might care more about false positive mistakes (precision) whereas others might care more about false negatives (recall).

Happy New Year everybody! If you got a fancy new Raspberry Pi 5 for Christmas, you might be wondering what to do with it. Well, we have a suggestion: build a computer vision application with it! And we have all the tools you need to get started.

Raspberry Pi offers a great platform for computer vision (CV), ranging from home hobby projects to serious industrial applications. However, setting up a Raspberry Pi for computer vision can be a time-consuming process. Groundlight Pi-Gen, simplifies the setup process by providing ready-to-use OS images for Raspberry Pi.

Want to get the best chance of success from your new Groundlight detectors? Here are five suggestions from the Groundlight science team that can help you get the best performance possible.

Come at it from the point of view of making answering your image query question as easy as possible. Pretend you’re explaining the task to a novice. What would you need to do to set them up for success?

At Groundlight, we continue to build infrastructure that allows our customers to easily use computer vision without a pre-existing dataset for industrial inspection, retail analytics, mobile robotics, and much more. We've built many features towards the goal of declarative computer vision, and today we are excited to introduce FrameGrab, a Python library designed to make it easy to grab frames from cameras or streams.

FrameGrab supports generic USB cameras, RTSP streams, Basler USB cameras, Basler GigE cameras, and Intel RealSense depth cameras.