Aquatic Litter Monitoring system uses RGB images and composites from Unmanned Aerial Systems (UAS; drones) cameras along with Convolutional neural networks (CNN) for precise litter quantification and characterisation. The system ingests, stores, analyses and processes litter floating at surface waters in seas, rivers and lakes, and lying at beaches and shores and eventually delivers standardised classified litter data sets, which are fit for environmental management and indicators. With fine-tuned processing methodology, quality-checked training dataset of litter and the adoption of standardised protocols, the system is operational in the iMagine AI platform, providing valuable analysis to local stakeholders and clean-up operations in multiple countries. Comprehensive guidance and training materials are also available to facilitate the adoption of drone-based litter observation and image processing. Resulting litter data sets are important for environmental management, cleaning operations, and contributing to indicators such as the EU Marine Strategy Framework Directive (MSFD), the EU Water Framework Directive (WFD) and the EU Green Deal.
Addressing target audiences and expressing needs
- To raise awareness and possibly influence policy
- Use of research Infrastructure
- Collaboration
We want to further develop the service and model by using data from different countries, different environmental conditions, litter types and distributions. More diverse the data, more holistic the model, output and conclusions.
- Public or private funding institutions
- EU and Member State Policy-makers
- Research and Technology Organisations
R&D, Technology and Innovation aspects
The system is fully operational, offering the ability to analyse drone footage with a two-stage tile-wise classification approach. In future along with a broader database of images from other regions, the use of different AI methods could be employed to implement a finer division of waste categories as labels in order to further pursue the use of established waste categories.
The developed pipeline will be installable on the iMagine platform or on other hosted solutions or on personal computers, hence enabling deployment at various scales.
See the explanation for scalability.
See the explanation for scalability.
