Accessibility navigation


AI enabled bio waste contamination-scanner

Stahl, F. ORCID: https://orcid.org/0000-0002-4860-0203, Ferdinand, O., Nolle, L., Pehlken, A. and Zielinski, O. (2021) AI enabled bio waste contamination-scanner. In: AI-2021 Forty-first SGAI International Conference on Artificial Intelligence, 14-16 DEC 2021, Cambridge, England. (In Press)

[img] Text - Accepted Version
· Restricted to Repository staff only
· The Copyright of this document has not been checked yet. This may affect its availability.

2MB

It is advisable to refer to the publisher's version if you intend to cite from this work. See Guidance on citing.

Official URL: http://bcs-sgai.org/ai2021/

Abstract/Summary

In Germany Bio waste is collected in separate garbage bins from households in the municipalities (e.g. garden waste, kitchen waste, etc.) and composted. The end result is humus, which is finally fed back into agriculture and closes the organic materials cycle. Waste must be inspected for non-biological contaminants prior to composting, as these can compromise the composting process and damage screening equipment at the recycling facility. Undetected contaminants affect the quality of the humus and can lead to contaminants re-entering the food chain through agriculture. The paper presents a feasibility study of an automatic bio waste Contamination-Scanner aiming to catch contamination early in the recycling process. Image data of bio waste contamination has been collected from a recycling facility. These images were used to design, train and evaluate two Convolutional Neural Networks (CNNs) aimed at detecting contaminants during bio waste collection. One CNN was trained on RGB and the other on greyscale images. The results show an initial surface scan can detect contamination with an accuracy of up to 86\ and could form part of a holistic detector attached to bin lorries.

Item Type:Conference or Workshop Item (Paper)
Refereed:Yes
Divisions:Science > School of Mathematical, Physical and Computational Sciences > Department of Computer Science
ID Code:100390
Publisher:Springer LNAI

University Staff: Request a correction | Centaur Editors: Update this record

Page navigation