Improving labelling speed is our core goal. By creating quicker workflows and giving you access to more effective tools, our tests shows that all labelling tasks will be significanlty faster with Hasty.
It doesn’t matter if you’re doing quick-and-dirty bounding boxes or pixel perfect masks. We have the tools you need.
To further reduce time spent on annotation, we are using active learning. After you have labelled the first couple of images, our active learning-component will start to suggest new labels as you go.
PASCAL VOC, XML, JPG, EFG. Long story short, everyhing you would want to import can be imported into Hasty. And when you are done labelling, you can export your project to TensorFlow, Keras, or simply as JSON. It's your choice.
No more local installations that force you to spend hours pouring over cryptic documentation. No more emails telling your colleagues to switch to the new release.
Although we can’t release all of Hasty as open source, we are planning to release the core annotation tool at some point in 2019. Furthermore, we are planning to keep core functionalities free to use for all from first release to infinity.