The idea started with the desire to implement a way to throw trash away efficiently and in an environmentally friendly way. Sometimes, it is hard to know what bin trash might go to due to time or carelessness. Even though there are designated spots to throw different type of garbage, it is still not 100% reliable as you are relying on the human to make the decision correctly. We thought of making it easier for the human to throw trash by putting it on a platform and having an AI make the decision for them. Basically, a weight sensor will activate a camera that will take a picture of the object and run through its database to see which category the trash belongs in. There is already a database containing a lot of pictures of trash, but that database can constantly grow as more pictures are taken.
We think this will be a good way to reduce the difficulty in separating trash after it's taken to the dump sites, which should definitely make a positive impact on the environment. The device can be small enough and inexpensive enough at one point that it can be implemented everywhere.
We used azure custom vision to do the image analysis and image data storage, the telus iot starter kit for giving sensor data to the azure iot hub, and an arduino to control the motor that switches between plastic trash and tin can trash.