I am going to reject this pull request for a number of reasons:
It is only possible to create an instance of Visual Recognition in the US South and UK Regions, where the default endpoint is used.
Any implementation needs to be consistent with how endpoint setting has been implemented in other nodes, eg. Conversation, Discovery, STT etc.
Dedicated Cloud instances of Visual Recognition use a userid / password rather than an API key, so an implementation would need to be flexible. The reason this hasn't been implemented yet, is that I am not able to instantiate an instance of node-red in a dedicated instance, so can not test this out.
All committers have signed the CLA.