Pl@ntNet self-supervised vision models trained on large-scale plant image datasets for transferable visual representations (output=embeddings).