While brainstorming ideas, the environment was a topic that we wanted to focus on, but the areas of interest overwhelmed us. However, as we waited in line for lunch, we witnessed Styrofoam plates being thrown into paper and glass labelled bins, we soon found our topic. Styrofoam plates, among other materials such as food wrappers and ceramics, cannot be recycled contrary to popular belief. To expand on this idea, we also wanted to inform people that items such as food scraps should be composted, along with sharing that hazardous waste, electronics, and bulky items should be brought to special facilities.
We understand that if people do not make the small steps towards environmental change, there will be large consequences that could have been avoided to help our communities. We need everyone in society overall to put in effort for change, so we developed an easy to use app that is also accessible. It is able to educate people on how to make informed decisions on where to throw their waste in a matter of seconds--which can make a long lasting impact.
What it does
Our application allows the user to take a picture of their trash/waste to determine what type of waste product it is in a matter of seconds. It gives the user the possible types of waste along with their percentages to predict the best way of disposal. Additionally, there are clickable icons of each specified means of disposal, that provides information about each type of waste and where to dispose them.
How we built it
We created a Clarifai API custom model to predict what type of trash each waste item is, along with their percentage accuracy. We used react-native to design and customize the app. Also, we used expo which is needed to test and run our application for troubleshooting. The model along with our react-native design came together to make a user-friendly experience.
Challenges we ran into
This is our first time using Clarifai and react-native, and we often had to troubleshoot our navigation. We decided to custom make our model, and had to work through our issues about API keys, workflow, and different IDs while learning about the new react-native language. Additionally, the predictions that were displayed defaulted to the general model instead of our custom model which forced us to spend countless hours working through the code to direct a new ID.
Accomplishments that we're proud of
Learning a new API along with an almost entirely new language was definitely a challenge for us, however because of our persistence, we were able to learn all the content in less than 24 hours. Seeing our work finally come together made the whole process worth the frustration, and we are glad that we stuck with our original plan instead of switching gears when we felt like giving up.
What we learned
What's next for BinPredicter
In the future, we would like to improve the accuracy of our prediction model and publish it as its own app. In addition, we want it to be able to identify different objects in the same picture and which bins they belong in, even if they do not belong in the same one. We would also like to incorporate location features to locate nearby waste facilities. Overall, we hope that our effort put into the app makes a difference in our community and the environment to end this pressing social issue.