Spotting the difference: towards fully-automated population monitoring of African penguins Spheniscus demersus

摘要: Placing external monitoring devices onto seabirds can have deleterious effects on wel- fare and performance, even the most benign marking identification methods return sparse population data at a huge time effort cost. Consequently, there is growing interest in that minimise disturbance but still allow robust monitoring. We developed com- puter vision system automatically creates unique biometric identifier for individual adult African penguins Spheniscus demersus using natural markings chest plumage matches this against database. tested non-invasive field Robben Island, South Africa. False identifications of detected occurred less than 1 10 000 comparisons (n = 73 600, genuine acceptance rate 96.7%) to known individuals. The capacity was estimated be above 13% birds passed camera 1453). A significant increase lower bound recorded under favourable conditions. conclude suitable species: demonstrated sensitivity parable computer-aided animal systems literature. full deployment would identify more possible with complete exploitation cur- rent levels flipper banding Island. Our study illustrates potential fully-automated, non-invasive, wild animals.