Tomra introduces deep-learning sorting solution
Sorting solutions provider Tomra Recycling unveiled advancements in plastic sorting technology. At a special event held yesterday in Mülheim-Kärlich, Germany, the company introduced three innovative applications aimed at separating food-grade from non-food-grade plastics, with a special emphasis on PET, PP, and HDPE. These solutions leverage deep learning, a subset of AI, to enhance sorting efficiency and accuracy.
Tomra’s investment in Gain, a deep learning-based sorting add-on for their Autosort units, has enabled the rapid and efficient separation of food-grade plastics from non-food-grade ones for PET, PP, and HDPE on a large scale. This achievement addresses a significant industry challenge, as distinguishing between food and non-food packaging materials has been difficult due to their visual similarities.
The newly launched Gain next technology, a rebranded version of Gain, further enhances Tomra’s Autosort units’ sorting performance. By combining traditional near-infrared, visual spectrometry or other sensors with deep learning technology, the company has developed a highly accurate solution capable of achieving purity levels exceeding 95% in packaging applications.
In addition to food-grade sorting, Tomra has introduced two non-food applications: one for deinking paper to produce cleaner paper streams and another for PET cleaning, ensuring higher purity PET bottle streams. This advancement supports bottle-to-bottle quality, a critical aspect of recycling PET materials.
