Smart Waste Sorter

The problem of sorting trash is a big problem. Most people do not know if a piece of trash goes into recycling or compost, so they throw it in the regular trash which goes straight to the landfill. What they do not know is that when we add to landfills we are also adding to global warming. So to solve this problem my team, the Recycle Squad, and I built a Smart Waste Sorter for the Technovation challenge. This device detects if a piece of trash you are throwing is going in the right bin. If you aren’t throwing that piece of trash in the right bin then it beeps to let you know. However, many items go into the recycling, compost or trash categories, so we first started with the device detecting if the piece of trash was a bottle, piece of styrofoam, flimsy plastic, or if the image was of something that goes into the compost along with batteries. Then we continued to work so the device would detect if the piece of trash goes into the recycling, compost, or the trash bin.

How do you build a Smart Waste Sorter?

There were two steps to this project. The first step to building this device was building an Artificial Intelligence (AI) service that can read images and predict where that piece of trash belongs. The second part was building the hardware. The first step is to train the AI to do this task. For this, we needed to collect lots of example images that represent the three categories we wanted the AI to predict. There are resources like Kaggle, that have datasets for people to use, but we thought that we would use images off of google and take pictures of our own to build the database. Our dataset contained over 500 images, that we collected over a period of months. As we collected these images we also saved them to a google drive so the whole team had access to the images.

The image on the left is one picture that we got off of google and the one on the right is a picture of a piece of trash we took at home.

After collecting a lot of images, the next step was to train an AI to do image classification. The residual neural network is a popular image classification algorithm that is known to have high performance. We used it in Navigator (it’s a free tool available at this website) to train this image classifier. Navigator provides you with a REST API that will accept any image as an input and provide you with a prediction. The cool thing about this is that all you need is an internet connection and image that you want to predict. You can use it anywhere!

The Hardware

Once we got a pretty good accuracy for the AI, we worked on the second part of the project which was the hardware aspect. The Smart Waste Sorter has a motion sensor that triggers a camera to take a picture when someone is throwing a piece of trash. It then sends the photo to the AI service, so the AI can detect if the piece of trash is compost, recycling, or regular trash. It then alerts the person if they are throwing the piece of trash in the wrong bin.

We believe that by installing such a device it benefits not only the community but also the environment around us. We also hope to educate our community on what to do when it comes to throwing the trash.

https://www.youtube.com/watch?v=_HWg3pvM9tU