Submitted Ideas

Google any ocean species’ picture

Ocean species are disappearing but we have hardly any information on where they are in the first place. Identifying species is hard and beyond the expertise of most people. We need to empower citizen naturalists to collect data that will inform us on the species in our oceans.

Google and other entities are using Deep Learning models to automatically identify things and people in images. There are thousands, maybe millions, of images of ocean species on the internet now, in databases like FishBase and a multitude of other sources. We propose to mine these images and create a hierarchical model that can automatically identify a species from an image. Expert users and organizations like museums would be able to upload images of identified species into the system to improve the model.