Advancements in AI for automated marine litter detection and management

Presentation by Dr Fereshteh Nayyeri (CSIRO)

Abstract

Marine scientists utilize diverse methods to monitor macro-litter in aquatic environments. Monitoring floating litter poses a significant challenge as conventional identification methods demand labour-intensive visual recognition. Recent AI developments that use deep-learning-based litter detection offer promising cost-effective solutions. Our project utilizes AI technology with cameras over Sydney’s waterways to capture and analyse images of floating litter. Our project has developed computer vision models that can accurately identify and classify litter types and quantities flowing down our urban waterways. Our technology has the potential to reduce manual visual survey effort and improve data accuracy. Our approach enhances the efficiency of litter management and waterway health.