GPU-Optimized Sortation Software

Digimarc Corporation, a pioneer and global expert in digital watermarking technologies, announced the release of its new Digimarc Recycle sortation software. This technological advancement reduces the cost of Digimarc Recycle-compliant hardware by nearly 50%, significantly lowering the barrier of entry for recycling and waste sortation facilities around the world that are eager to deploy a more sophisticated and much-needed solution.

Digimarc Recycle addresses a critical industry need: it increases the precision and accuracy with which sorting machines can sort recycled material. By identifying digital watermarks on product packaging, the software consistently and accurately determines each item’s composition. Current technologies offer limited sortation parameters and are prone to misidentification, often resulting in cross-contaminated plastic that cannot be close-loop recycled. Next-gen sortation is essential to making a circular plastics economy viable. Digimarc Recycle accurately identifies plastic packaging during sortation, precisely determining the product to enable sorting with any desired level of granularity. This is achieved by linking covert digital watermarks applied to plastic packaging with a cloud-based repository of extensible product attributes—including packaging composition, food or non-food grade plastic, product variant, brand, SKU, and more.

This increase in sortation specificity is achieved using sorting machines that employ cameras and Digimarc Recycle software to detect recycled materials on moving belts. In the latest version of its Recycle sortation software, Digimarc has redesigned its detector operations to work with Graphics Processing Units (GPUs) that are increasingly used in AI applications.

“In redesigning our detector software operations for GPUs, we’re harnessing the parallel processing capabilities and the performance and cost advantages they provide,” says Ravi Sharma, Digimarc’s Vice President for Research and Development. “As a result, we’ve reduced the hardware needed for our detection technology, thereby reducing the compute cost of Digimarc’s detection module by up to 70%, while achieving the same performance. It’s a win-win for recycling facilities and the industry, as our latest release increases the ROI for all recycling ecosystem participants.”

Digital watermarking is the only advanced sortation solution capable of accurately detecting and sorting distinct plastic varieties, including food, cosmetic, and hygiene grade plastics. The technology works for all form factors including rigid and flexible packaging. This makes it a more holistic and immediate solution to drive circularity. In addition, the SKU-level, product-variant level, or even item-level specificity that digital watermarking provides is essential for meeting traceability requirements set forth by the Packaging and Packaging Waste Regulation (PPWR).

“The problem of plastic pollution is growing exponentially,” says Riley McCormack, Digimarc’s CEO. “We need solutions that are profitable, effective, and can revolutionize the recycling economy now–not in a few years when the problem will be dramatically worse. Digimarc Recycle is ready and available today.”

Digimarc’s latest version of the Recycle software is fully operational in the Hündgen material recovery facility in Germany, installed in a sorting machine created by Pellenc ST, a leading manufacturer of optical sorting machines. Within a day of the module being functional, tens of thousands of Digimarc digital watermarked items were identified at the facility, demonstrating that this technology is not only available, but also effective at scale.

“Achieving such a significant reduction in cost while maintaining efficacy is a key success factor for implementation of the technology in the recycling industry,” says Antoine Bourely, co-founder and Chief Scientific Officer at Pellenc ST. “We are eagerly looking forward to the use and validation of this latest version of Digimarc’s Recycle software using GPUs in industrial conditions.”