I have not used any pre-trained model with Azure AI Vision; we are using only a customized model. For Azure AI Vision, we wanted a complete pre-trained model. Actually, we did not find any pre-trained model for a specific domain, so we are building from scratch. I know it is difficult to build it from scratch. For example, in a legal department, we are using the Iron Legal BERT model, a pre-trained model for law. We are expecting similar pre-trained models in medical or emergency services so that we can improve. I see Med-Image-Insight, which is one application being used. It is already open, but we want to see it more in different domains as well. I want a complete pre-trained model inbuilt with Azure AI Vision so that it can help in other domains as well. Then I can directly consume the API and solve the problem. For now, we are using customized code, which may be vulnerable to security issues, and we do not have access to datasets in certain domains. If they can come up with a complete customized or complete pre-trained model, I would say ten out of ten is achievable.
