Enhancing citizen science campaigns through artificial intelligence methods

Abstract

Many studies show that citizen science initiatives are a very useful tool for data collection and a way to overcome limitations of time and resources. The first part of this study focuses on the creation of a national database that formally documents all the marine alien species found within Maltese waters, including the sighting date and time, location, photographic evidence, name of the species, and other information. The second part is dedicated to the applicability of machine learning methods for marine species identification. Hundreds of photos that were submitted to the “spot the Alien” initiative were used to train a region-based, convolution neural network. The main aim was to develop a model that can classify and distinguish between the eight most recorded marine alien species within Maltese waters: 'Abudefdud saxalitis', 'Acanthus monroviae', 'Staphanolepis diaspros', 'Portunus segnis', 'Seriola fasciata', 'Siganus luridus', 'Aplysia dactyomela', 'Lagocephalus sceleratus'. A number of metrics were calculated to quantify the reliability of the model. The use of the model can reduce or even eliminate, the need for human expert intervention in validating citizen science reports and will provide prompt feedback to the citizen scientist submitting the report. In addition, a web portal with visualization tools to help display the information in database, was implemented. This point of reference allows users to upload images of marine alien species, which can automatically be classified by the R-CNN. The reports will continue to populate the national database. This work will enhance citizen science campaigns that have been running for several years and that target the monitoring of the influx of alien fish into Maltese waters.