Artificial Intelligence (AI, using machine learning and neural networks) has made amazing strides, notably in recognising human faces. It is also being used to identify patterns on photographs of individual patterns on whale fins and flukes, and whale shark markings. It is revolutionary in helping citizens learn how to identify species (such as in iNaturalist) and is poised to enable automation of biodiversity monitoring through analysis of videos and photographs. The use of images has the added benefit of having minimal disturbance to biodiversity (nothing being killed) and images can be archived for future research. Here are some examples I compiled recently. Please add more to the comments box below.

iNaturalist uses AI to put a taxonomic name (species or higher level) against one species per image. Volunteer experts help confirm identifications and images with >100 confirmed identifications are used to train the AI. Images are archived. Data is automatically published to GBIF.

MerlinID uses AI on images and observation context to help identify birds.

Ecotaxa is a system for Identification Plankton images using AI.

www PIC is a plankton image archive with applications for AI training.

Linne Lens identifies multiple animals (especially fish) in real-time on videos from a smart-phone app. It can count and name fish and other species in videos and photographs.

CoralNet AI (deep neural networks) to annotate benthic images. It is in use for semi-automated annotation of benthic images of coral and rocky reefs.

FathomNet  MBARI system for training AI using expert knowledge to detect marine species (even detecting a species is present can be useful to save time watching many hours of video).

Squidle+  New Australian platform for marine image storage, mapping and annotation supported by Schmidt Ocean Foundation and Australian Integrated Marine Observing System has potential for community image storage, expert annotation and AI training.

VIAME is NOAA’s AI for video analysis, detects fish in images and videos. Fish are detected in videos which are then expert annotated.

Bisque (Bio-Image Semantic Query User Environment) stores, visualizes, organizes and analyze images in the cloud. is a specialist in environmental image analysis including having an app for species identification in Japan.

Examples of AI use

The “CORaiL platform” uses video to monitors growth of transplanted corals on concrete reefs near Philippines’ Pangatalan Island

Detection of plastic floating in the ocean from satellite images


Print Friendly, PDF & Email