Hasty GmbH is a vision-based AI startup from Berlin. We are a product of WATTx, Viessmann GmbH’s company builder.
What types of images can I annotate with Hasty?
At the moment, the tool can only handle 2d images formatted in jpeg, png, bmp, tiff, or GIF.
Currently Hasty is free - will it stay that way?
Some parts of Hasty will always be free to use. For example, all non-AI annotation tools will always be free to use. We also promise that we will never limit the amount of users permitted for a free project at any given time.
Beyond that, we also commit to give everyone that wants to use Hasty for free 3000 credits per month (to a value of 30€) so that all users can experiment with our AI tools without paying first.
However, Hasty will not stay 100% free. Starting in February 2020 we will go to be a freemium tool. In short, everything you can do with the tool today will stay free, but we will create new functionality that only paying users will be able to use.
We know that you may already have started annotating your data and do not want to duplicate the work. That's why we've made it possible for you to upload annotations into Hasty. More information on importing annotations can be found here. On request, we can also help you switch to Hasty.
You can request alternative hosting methods by contacting us at [email protected] Note that these special options charge much higher fees than the standard prices of our SaaS solution as they are intended for enterprise customers.
Will you have an API that I can use?
We will have an API that will be available for paying users in February 2020.
We can offer you data labelling services as we have established partnerships with several annotation service providers. We have partners of all sizes. So if you need to label 10 images or 10 million images, we can help you connect and realize the project. As all data labelling will be done using Hasty, we can also offer advantageous prices compared with other services.
Your projects and image data are only visible to you and other users that you invite. We take information security very seriously and will never share team information or image data with third parties. We also adhere to the strictest security standards to ensure the security of your information. You can read more about our data policy in our Terms of Service.
If you have a question that is not covered in our documentation, or want to suggest a new feature, or just want to know more about what we have planned for the future - send us an email to [email protected].
Many other applications today use various technological methods to improve the quality of images (anti-aliasing and pixel smoothing). This creates a better to look at image for humans, but doing this type of post-processing can be dangerous when working with machine learning, as these techniques changes the composition of an image. Therefore, we decided to turn these post-processing technologies off. It might make images seem more pixelated, but it also means that images will be presented in their original form, which is what will be used for machine learning later. In essence, it looks slightly worse but allows for better annotations.
Hasty ignores Exif metadata like rotational instructions. Our reasons for doing so are two-fold. The first is that browser support for understanding and acting on this type of data is spotty, and the range of browsers used to work on Hasty is quite large. The second is supporting Exif rotational data might be more of a bug than a feature, as many users are unaware that their image contains it. It can lead to negative outcomes in the steps following annotation when users build models using annotations where the coordinates are mixed up just because of Exif.
If you have images that you know contain Exif data, and you want that rotation to be respected, please convert the images before uploading them to Hasty.