Can machine vision help save you money on your utilities? We think so. More importantly it can help reduce stress on recycling facilities.

Using machine vision we can detect problematic materials that routinely are mistakenly put into the recycling bin, such as thin film plastics. These items can not be recycled and cause a lot of strain on the machinery when attempting to sort items that can be recycled.

By incentivizing customers to sort their items before it reaches the sorting facility, we can increase the overall throughput and efficiency of our recycling facilities and subsequently increase their yield and productivity.

Here’s how.

Synapse Software Development Engineering Program Lead, Jonathan Ross, talks about the power of machine vision and how it could be used to sort recycled items and reduce strain on recycling facilities.


00:00:00:01 – 00:00:29:54: There are some staggering statistics out there that say that only roughly 9% of plastics that we use end up being recycled. Roughly 50% of that plastic ends up in a landfill. And of the remaining, some of it is incinerated and the rest is disposed of using uncontrolled methods. Now, there are many contributing factors to this problem. In some places, recycling facilities aren’t available, but when they are, we’ve learned that sorting is a huge challenge at the recycling facility.

00:00:30:32 – 00:00:49:33: Helping conserve the environment is really important to us here at Synapse, and we asked ourselves what can be done to alleviate this sorting process and how can we use technology to help alleviate the problem? And here’s what we’ve come up with. It would be possible to take footage at the home as the waste is transferred from the home to the collection vehicle.

00:00:49:46 – 00:01:19:10: We replicated the scenario in our lab using some simulated waste and some plastic bottles dumped from atop a ladder. We then took that video and broke it down into the component images and used a pre-trained network to detect plastic bottles. Using these detections, we annotated those images and recombine them into the video that you see here. In addition to detecting recyclables, we can use this technique to detect things like thin film plastics that can really inhibit the sorting process.

00:01:19:46 – 00:01:40:42: Taking this problem further, we can envision collection infrastructure, providing feedback directly to the users about what items are going into the wrong bin and potentially offering rewards to the customers who sort their waste and recycling properly. This ensures that the items that we want to recover don’t end up in the landfill and helps reduce stress on the sorting facility.

00:01:41:06 – 00:01:53:33: Now, there’s a lot more to this problem. So if you’ve got some ideas about what we can do to expand upon this, leave some comments down below and together we can help make this world a better place for years to come.