Detect Edge

Feb 15, 2026

In my last post, I talked about the “data bottleneck” in Edge AI. If you’ve ever tried to train a time series model for an embedded device, you likely know the struggle: the tools for training are amazing, but the tools for actually collecting and labeling that data in the real world are… lacking.

I often just want to test if there is any viability in a particular sensing application I’ve thought of, but getting to the point of training requires a good bit of effort to collect enough data. I got tired of cobbling together Python scripts and CSVs and carrying around my laptop, so I built the tool I’ve been wanting.

I’m excited to share Detect Edge.

Detect Edge is a data collection and labeling mobile application for time-series data captured from the Nordic Thingy:53. Here is how it changes the workflow:

• Real-time Visualization: You can view the sensor stream in a graph, which is critical for making sure you are actually capturing the event you’re after correctly.
• Edge Impulse Integration: After collecting your data, you can upload it to a project in a few seconds using an API key.

This dramatically reduces the amount of time spent experimenting on new applications. Instead of hours of data management, you can move from an idea to a trained model much faster.

Getting Started

I’d love for you to take it for a spin. It’s now available on both the iOS App Store and Google Play. You can find a distributor for the Thingy:53 here, if you don’t already have one.

My ultimate goal is to enable anyone to train and deploy a model based sensing device for an application in minutes, not days or weeks. I’ll be sharing end-to-end demonstration projects on detectlabs.com, with models you can easily test out yourself.

If you’re working on an Edge AI project, or interested in learning more about them, please reach out to me via one of the socials below.