Two basic features missing

Hello there!

I’ve been evaluating Hasty for my company use but I feel like there are two basic features missing:

  1. Big cross: When annotating a bounding box, you need that the “cross pointer” fills the whole screen. With a small one you can’t really judge where to put the two points.
  2. Image brightness adjustment.

There are also other two features that would be helpfull for us:

  • A slider for annotation opacity.
  • Select which classes are going to be used for training (for the AI tools): we use a custom annotation procedure where we have to annotate some parts of the image but these annotations would be wrongfully interpreted by the AI tools.

Best regards,

Hi Caio,

thanks for reaching out. Feedback like this is always super valuable for us! Your name sounds Brazilian, am I right about this?

I’ll reply to your points in chronological order:

  • We’re actually refactoring the front-end of our annotation environment at the moment. The new version will have such a cross when you create labels. It’s still under closed beta, but if you’d like to try it out, drop me an email at tobias@hasty.ai. Then I’ll make sure to give you access to the beta.
  • Adjusting the brightness of images is a very good point. I forwarded this to our product team. Until we have this implemented, You could do this easily in pre-processing, though. Of course, this would mean some manual work for you, but with the time our assistants save you, you still should be faster than using other tools :wink:
  • You can adjust the opacity for annotations already. When you click on the pen next to a class name, you can edit the class’ color and its opacity when you click on the arrow next to the rectangle:

image

  • Also, your last point is a good one. I forwarded it to our product team as well. But again, you could solve the issue with a workaround: You could create two different projects; one for the classes which our assistants should be trained for, and one for the other ones. As long as the filenames are the same in both projects, you could merge the two JSONs you get after exporting your annotations. You’ll need to do some manual post-processing, but it still using Hasty should save you some time overall, taking the speed you gain from our assistants into consideration.

Generally speaking, how are our assistants working out for you? Are they producing good results? Did you pick up that they’re constantly retrained and improve over time?

Thanks again for your feedback, and I hope that my answer could help you.

Have a good Sunday,
Tobias

Thanks for the detailed answer and sugestions.

Indeed I’m Brazilian.

We have used the assistant before in a smaller, segmentation task and it did help us, especially for highlinghting what images we needed to annotate.

Best regards,

Que beleza! Meu pai é Brasileiro. Sadly, my Portuguese is not so good, so I’m going to stick to English here.

You’re very welcome. I hope that I could help you. We’d be more than happy if you’d come back and decide to use our tool on project. If you have any questions, feel free to reach out any time. I also could give you a tour through Hasty. Would you be interested in jumping on a call? There are one or two tricks I can show you that you only know when you work here.

Just drop me an email.
:v: Tobias