Carton Council opens image library to sharpen AI carton sorting at MRFs

Cartons rarely make up more than 1% of inbound tons at U.S. materials recovery facilities, and that scarcity now has an institutional answer. The Carton Council has built an open-source image library stocked with photos of food and beverage cartons as they actually arrive at MRFs — torn, dirty and squashed — to train the AI-powered recognition and sorting systems that increasingly decide what gets recovered.

The library addresses a structural problem in machine learning for sorting. Because cartons are a thin slice of the inbound stream, they appear less often in the training datasets that teach robots and vision systems to identify materials. Fewer images mean weaker identification and lower sorting reliability. The Carton Council's communications manager, Brendon Holland, framed the gap plainly: the library exists to supplement datasets where cartons are underrepresented.

Who is using the data?

Glacier, a robotics developer, is the first company to draw on the library to train its machines. Margaret Link, a Glacier spokesperson, said AI-enabled robotics are often trained on billions of images and that "more is more" when the images reflect real MRF conditions. Flattened, twisted cartons across a range of brands and sizes beat shiny retail product photos for training value.

"When we don't have the entire picture of the real world, it means that things are getting missed. It means that the robot can't do its job as effectively," Link said. "That's why we are crazy about trying to increase the breadth of the data that feeds our machine learning."

The library supplements Glacier's existing in-house dataset. Other companies have already requested access. Holland said the Carton Council does not endorse any specific company or technology and will work with any equipment provider or operator that wants to contribute images or test equipment against the library.

Where did the idea come from?

The project grew from a casual conversation last year between Carton Council and Glacier. A Glacier representative noted the company could use a broader set of carton images to fine-tune operations at a California MRF where it had just installed a new robot — an upgrade the Carton Council had partially funded with a grant.

The Council sent photos from a bale audit then underway, covering material from roughly 10 different MRFs. The exchange suggested a bigger play.

"But then we thought, 'Well, why not create an open-source database to improve recycling for everyone?'" Holland said. "Everyone has their own in-house training data, but if we also can work with MRFs to get images that are representative from recovery screens that equipment providers don't often see, like regional differences, we can help improve the system across the board."

What is the regulatory backdrop?

California supplies the sharpest deadline pressure. The industry has pushed carton collection and recycling rates upward to comply with SB 343, the state's "truth in labeling" law. After months of scrutiny of recycling rates under that statute, cartons are once again being sorted for recycling at several California MRFs.

Holland argued that precise sorting matters nationwide beyond carton end markets. Better carton identification also cleans residue streams and cuts contamination in jurisdictions where curbside programs do not accept cartons at all.

"A 1% increase in recovery might not seem like that huge a lever, but 1% reduced contamination would be huge," he said. "We're looking at it on the side of cleaning up bales so that markets are better able to process them. That, to us, is worth a considerable amount of effort to make it easier to recycle material that is collected."

The next milestone to watch is adoption. The library's value compounds as more MRFs contribute images from recovery screens and more equipment providers test against them. Glacier is first; the queue of pending requests will determine whether an open-source dataset becomes industry infrastructure or a single vendor's training supplement.