My team wants to manually annotate 30000 images using image.manual so does prodigy allow multiple user access? If yes, how to go about it? In such a scenario, will we be shown different images? if not, will there be multiple copies of the same image in the output and what if an image is given different labels by my team members, how does image.manual deal with it?
When you run a recipe, Prodigy will start up the server on the given host and port and stream in the data. When you start another server on a different host and port, it will do the exact same thing.
If you're working with multiple annotators, you sometimes want to have everyone annotate the exact same data and then compare the results. In other cases, it makes more sense to distribute the data between your annotators. When you come across conflicting annotations, you'll later have to decide how you want to use them for training later on. For example, you could export all data, extract the conflicting annotations and then reannotate them yourself – or have an automated process that only includes labels if X% of your annotators agree on them.
We're currently working on the Prodigy Annotation Manager, which will be an extension product to Prodigy and will allow users to manage large annotation teams, create bigger datasets and assess annotator performance.
If you search the forum for "multiple annotators", you'll find DIY solutions and examples of using Prodigy with multiple annotators in different scenarios.
There's also this open-source extension developed by a fellow Prodigy user and forum member: