Due to it being a new and fast-changing space, it's difficult to find reliable information on the different platforms/companies/services available to outsource the labelling of training datasets.
Which platforms / services do data scientists actually use to have their data labelled, and what are the pros/cons/limitations of each?
What I know so far
I have so far used freelancers to label data, which has been successful except that a specialist platform/service would have:
- greater specialisation (people who do this task regularly will be better and more reliable at it)
- hopefully automate some of the setup processes (in the ideal world, a perfectly programatic interface could involve minimal/no face-to-face or even voice meetings, just some information (text/images) on what is required and the dataset to be labelled)
- improved integrity (some, but certainly not all datasets can be given to freelancers)
I found the following providers online, but cannot say the pros/cons/limits of them (as I haven't used them)