You could have an LLM analyze pictures of each card to extract the attributes you want, eg. to json, then ask it to write code that will input it to your librar

You could have an LLM analyze pictures of each card to extract the attributes you want, eg. to json, then ask it to write code that will input it to your library. The intermediate step of json gives a chance to check and clean the data before upload.