Ask HN: How do you annotate your data?
Hey HN,
I'm curious how folks annotate their data to be used in ML pipelines (NLP/Vision) and what their experiences have been. Production/applied examples with teams of annotators would be great.
From folks I've spoken with these seem to be the most common ways:
* Custom Solution * Prodigy * Dataturks * Lighttag
I'm curious if there are more latent examples leveraging your users such as Google's image captcha. I built 2 annotation tools for the company I work for. one is web based, the other is a c++ app that can directly annotate videos. the web based one is abandoned, because I found it is difficult to do frame level navigation. Is there a reason you chose to build your own instead of using an off the shelf solution?