Machine Learning in Recycling Industry
The World Counts on its website stated that each year we dump a massive 2.12 billion tons of waste with over 50 different types of waste listed. By 2050, global waste is expected to increase to 3.40 billion tonnes, more than double the population growth over the same period according to the World Bank Group.
Do we just stand there and watch as the world is filled with waste? Of course not, right? Because back in 1970, we have been encouraged by environmentalists to embrace the “Reduce, Reuse, Recycle” as a call to action to save the environment. Now, along with the advancement of technology especially machine learning, we can use the “machine” to help us in identifying the waste that later on will be Reduced, Reused, and Recycled.
One of the ways to implement machine learning in the recycling industry is to use what is called object recognition. Simply put, object recognition is a technique in computer vision (a field of artificial intelligence) for identifying objects in images or videos.
Nowadays, a lot of companies have tried to use object recognition in the recycling industry. Superbin for example, a company based in Korea, has been working with Tictag for the past year to annotate their data which will later be used to train their machine learning model. To learn more about data annotation, you can click on this article: What is Data Annotation?
So what are the benefits of using machine learning for the recycling industry?
- Reduce reliance on manual labor
One of the interesting ways Superbin does recycling waste is that they use some sort of vending machine with a camera inside of it to help them recycle the waste. People who have waste in their hands such as plastic bottles can simply put their waste in the vending machine and the vending machine will automatically try to identify the waste. By using this, instead of using manual labor, Superbin can put as many vending machines as possible to help reduce the waste produced in Korea.
2. Faster in waste management
We know that machines, given enough data to train will become faster and smarter than humans and this is the case with sorting types of waste. There are a lot of waste types and some of them are non-recyclable goods such as plastic grocery bags, styrofoam, and broken or sharp objects. By using object recognition algorithms, the process of sorting types of waste can be significantly faster and the machine can also be updated each time there is a new type of waste.
3. Better quality control
Using machine learning will also ensure besides quicker waste management, but also more precise in terms of waste being picked out and sent to the correct type. This is also given enough resources for the model to train so that the machine is able to do such tremendous work.
This doesn't mean that there are no cons to using machine learning for recycling. There are some cons to this such as the reliance on the power source, how time-consuming it is to do data annotation (this is why Tictag is here btw), etc.
But aside from the cons, I believe that there are no cons that could stop us from advancing to a more reliable way to Reduce, Reuse, and Recycle and that way is using machine learning.
References:
