Just took a look through the demo and tried some labeling for myself. I manage a small team of labelers using other tools that are overly complex and pricey. Pros: "Label assist" is the killer feature for me, allowing for quick corrections to labeled images. The "auto-label" also is a huge time-saver. Simple UX without unneeded frills. Cons: No collaboration features yet.

Looks interesting! I can imagine that Dashup would save a lot of time in a data labeling workflow. What’s the difference between auto-labeling and label assist? Congrats on the launch! 🚀

Hi Product Hunt, Marco here. I'm a Computer Vision Researcher and have worked on several academic projects. I've always been frustrated by the existing Data Labelling offerings. They're either clunky to use, require self hosting, or lack basic automation tools. That's why I built Dashup! - a free tool to automate Data Labelling. Hopefully with Dashup MLEs can spend less time labelling their data, and more time training and deploying models. There are two automation tools in Dashup that I'm particularly excited about and have saved me a ton of time in previous projects: The Auto-Labeler: This is a model that uses previously labelled images as context to automatically suggest labels for new images. The Label-Assist: This tool will takes a very rough outline of an object and sharpens the edges to fit the object. It's like a super-powered version of Photoshop's magic wand. If you're interested please give the product a try! We welcome any feedback and feature requests!