According to foreign media reports, a new study shows that cutting-edge artificial intelligence may help completely change the way waste is recycled in the UK. Researchers from Hope University in Liverpool are developing a new low-cost classification system that can reduce the amount of solid waste entering landfills.
It is understood that this new method uses a system based on a “computer vision” camera, which can recognize every household item on the conveyor belt after training, and then instruct the robot to classify the items in front of them, rather than let the recycling center rely on human manual screening items.
Dr. Emanuele Lindo SECCO, of the school of mathematics, computer science and engineering at Hope University, said they had been able to prove that the method was effective and accurate.
It’s worth noting that he and his co-author, engineering and robotics tutor Karl Myers have also successfully produced this classification system for less than 100 pounds, making it possible to promote it all over the world.
Secco and Myers wrote in a new report: “due to rapid urbanization, population growth and industrialization, global solid waste pollution is rising sharply. Can we handle so much rubbish? The answer to this question is No. At present, we don’t have the ability to deal with more and more garbage, and we don’t have the ability to deal with the garbage we are recycling. Therefore, we must strive to simplify the waste classification process and strengthen the work of intelligent waste recovery, so as to further reduce the pressure on material recovery facilities. “
Secco is said to have used a cheap raspberry pie computer and combined it with a high-resolution camera. Through intelligent machine learning, raspberry pie computers are programmed to recognize five different types of waste paper, glass, plastic, metal and cardboard.
And it can do that because it has a database of 3500 different junk images, provided by Google images and a resource called trashnet.
Through the training of “transfer learning” — one of the key driving forces for the realization of “real artificial general intelligence” (AGI), the system can become more and more powerful in identifying garbage labels.
When the team talked about the overall accuracy, they said it could pass the test with a success rate of 92%, which made it a commercial application prospect.
Although Dr SECCO admits that the speed of the system needs to be further improved before it can really be put into use – thanks in large part to the processing power of raspberry pie, the basis for a new processing method has been determined.
SECCO added: “the world we live in is increasingly polluted by human made waste. Great progress has been made recently in recycling these wastes, but this will increase the pressure on already inefficient material recovery facilities. Therefore, more efforts must be made to improve the efficiency and reduce the cost of MRF. More specifically, computer vision (CV) research shows that embedded CV may be the answer to this and many other related questions. Especially now, because of advances in technology and software, it’s easier to get, easier to use, and more powerful. “