Sounds good, huh? Onboarding takes no more than five minutes, so let's go! 💥
Supervisely is free for non-commercial usage and requires no card info. Go to the signup page, choose login and password to create your personal account. You can sign in through the GitHub or Google Account too.
When you create an account we automatically create a new team for you that has the same name as your login. You are the only member right now — but you can invite your colleagues and work on the same data together!
Go to the import page and create a new project by uploading some data: images (.png, .jpg), videos (.mp4, .avi), 3D point clouds (.pcd) or even well known dataset formats like KITTY and PascalVOC.
We support such a variety of formats using plugins concept: and you can always add your own!
Go to the projects page to see your new project: a set of datasets (aka folders) that share the same settings, defined tags and so on. Click on a dataset to open corresponding annotation tool and create some polygons or bitmaps or tag your images.
Get some benefit from your datasets — train a new neural network or apply an existing one. By the way, we have a bunch of them in our model zoo: like UNet or DeepLab.
Before you start, you need to connect a computer with GPU on board to the platform, so that you can deploy new tasks right from the web interface.
Thankfully, it's super easy to do: just add a new agent at the cluster page and run the one-liner on your computer: that's it!
Now you know the basics, but there is so much more to explore! Data augmentation, consensus labeling, data statistics and visualization, Python SDK and API and even more.
A nice idea would be to learn some core concepts about Supervisely: