An intelligent platform to perform automated airborne wildlife census in the marine ecosystem using an ensemble of learning techniques and computer vision

Abstract

"The habitats of marine life, characteristics of species and the diverse mix of maritime industries around these habitats are of interest to many researchers, authorities and policymakers. Automated detection, location and monitoring of marine life, along with the industry around the habitats of this ecosystem, may be helpful to i) reveal current impacts, ii) model future possible ecological trends and iii) determine required policies accordingly leading to the reduced ecological footprint and increased sustainability. New automatic techniques are required in order to observe this large environment efficiently. Within this context, the aim of this study is to develop a novel platform in order to detect marine ecosystems and perform bio census in an automated manner, particularly for birds in regional surveys composed of aerial images. In this manner, a new non-parametric approach, WILDetect, has been built using an ensemble of supervised Machine Learning (ML) and Reinforcement Learning (RL) techniques. It employs several hybrid techniques to segment, split and count maritime species - in particular, birds - in order to perform automated censuses in a highly dynamic marine ecosystem. The efficacy of the proposed approach is demonstrated by experiments performed on 26 surveys which included Northern gannets (Morus bassanus), based on retrospective data analysis techniques. With this platform, gannets can be detected and split automatically with very high sensitivity (Se) (> 0.98) and specificity (Sp) (> 0.97) values. The experimental results suggest that similar autonomous techniques tailored for specific species may be helpful in performing marine bio censuses efficiently."