We use machine learning technology to identify various types and forms of waste (plastic, paper, glass, metal).
We use machine learning facilities provided by Google, namely https://teachablemachine.withgoogle.com. To obtain a model, we collect various types of waste to be used as a sample. the waste object is photographed and then given the class name according to its type, for example plastic001, paper001, glass001, metal001. The type name at the beginning of the class is used by the program to provide notification. We upload the results of this model train to the server and used as model reference to be linked to pictoblox,
In pictoblox we use a Machine Learning extension to classify images, a Human Body Detection extension to detect hands, and a Text to Speech extension to tell humans by voice.
We use the camera as an input tool for the system to recognize the garbage around it. The type of garbage that is on human hands will be detected and the system will notify humans by using a sound and an indicator light placed in one of the trash, the light cans will turn on which indicates that the waste must be disposed of in that place
Although the level of accuracy in recognizing waste is not yet optimal, we are optimistic that this solution can help solve the current waste problem, and the most important thing is the increasing awareness of waste management from young people like us.
Model Link Reference:
https://teachablemachine.withgoogle.com/models/A_rWpnYb8/