KumoVision is a set of open-source tools that covers the whole computer vision pipeline, from the first labeled image to a model running as a service. Everything you build with it stays yours: we build on the Hugging Face Hub, with MONAI for medical imaging, so you can take your models and run them without us.

The tools

🏷️ KumoLabel: label thousands of images at once

KumoLabel sorts your images by what's actually in them, so you label a whole cluster in one go instead of clicking through them one by one.

🎯 KumoTrack: annotate one frame instead of many

Mark an object once, and KumoTrack follows it through the rest of the video. You only step back in where it loses track.

🔗 KumoFlow: connect models into a pipeline, without code

Connect your models on screen and run them as one pipeline, with whatever you built in the other Kumo tools and with models from the Hub. Deploy it, and it runs as a service the rest of your software can use.

Get started

• 💻 Code: github.com/iits-consulting/kumo-vision
• 🤝 Want us to deploy it, integrate your models or set up the hardware? Write to projekte@iits-consulting.de
• 🏢 About us: iits-consulting.de

_{Made in the EU 🇪🇺}