At TDWG 2018, someone (from Google?) was interested in doing machine learning on images to find out what kind of thing they were (skull, dead animal, etc.) This is related to #118, since machine learning might help automate the process of assigning ac:subjectPart values.
This is not an actionable issue. The idea of facilitating machine learning is included in one of the submitted use cases for the task group, so there is no reason to maintain it as an issue here.
At TDWG 2018, someone (from Google?) was interested in doing machine learning on images to find out what kind of thing they were (skull, dead animal, etc.) This is related to #118, since machine learning might help automate the process of assigning ac:subjectPart values.