Multi-Spectral Segmentation with Synthesized Data for Refuse Sorting

Refuse sorting is a key technology to increase the recycling rate and reduce the growths of landfills worldwide. However, monitoring and parameterization of sorting facilities is still done in a mostly static fashion. This work combines multi-spectral imaging with deep learning based image recognition to monitor and dynamically optimize processes in sorting facilities.Our solution is capable of monitoring the sorting process remotely avoiding potentially harmful working conditions due to dust, bacteria, and fungal spores. Furthermore, the introduction of objective sorting performance measures enables informed decisions to improve the sorting parameters and react quicker to changes in the refuse composition.