Herbarium GB at the Department of Biology and Environmental Science, University of Gothenburg, has generated high-resolution images of c. 700000 vascular specimens collected during the past two centuries around the world. In order to make metadata (collection time and locality, collector, etc) available for text-based search, we plan a combined artificial intelligence(AI)/citizen science approach. In this first step, we intend to text a few freely available approaches AI approaches. Hespi uses NVIDIA's CUDA library and hence we request for NVIDIA GPU nodes.