Save the Earth

An android app applied machine-learning to classify captured recyclable and non-recyclable materials into a number of classifications

Comment

3 Categories to Apply

  • Best Social Impact App
  • Cleanest UI/UX
  • Best (World's) Health App

Inspiration

It is widely known that trash harms the Earth, which eventually harms the humanity by negative outcomes. People try to eliminate the amount of trash by recycling, but many people do not participate on recycling mostly because they do not know what to recycle, and they feel inconvenient to search for that information every single time they recycle. Also, since one-stream recycling (which basically allows people to recycle everything in one recycling bin), trashes (such as candy wrappers or glasses for facilities that do not accept glasses) also has been put into the recycling bins that damage expensive recycling machines. To prevent such happening and encourage more recycling in a right way, I developed a recycling app, which I named Save the Earth. As a motivator, I included daily goals to achieve and collections that show the statistical information of user recycling behaviors.

What it does

Users can take a picture of an object to find out whether it is recyclable or not. Users can also accomplish some goals by taking pictures to recycle as well as statistical information about their recycling behaviors. Users can also see collected photos and search a specific item to find out if it is recyclable.

How I built it

  1. Google Cloud & Google Colab I connected Google Cloud to run machine learning model in order to produce image classification. Each category for the image classification uses about 100 images to extract features.

  2. Android Studio (Java) I coded all the functionalities in Java and XML.

Challenges I ran into

  1. It was quite difficult to adapt machine learning model in Android Studio. I had to search hours to figure out how.

  2. I had issues to calculate some numbers. For instance, I wish to show a percentage of how much a user recycled items more or less, but because there are many other numerical variables, I got confused when I was writing my formula and took awhile to figure that out.

  3. I hoped to offer weekly and daily goals, but weekly goals did not work out well, so I removed them.

Accomplishments that I'm proud of

This is my first android app that I fully complete all functionalities I designed!

What I learned

  1. Machine Learning I learned about fundamentals of image classification and other related classifications, such as object detection or segmentation. I also learned about some of pythons that are used to generate machine learning models for image classification.

  2. Android Studio I learned to use many new functions, such as SharedPreferences.

  3. GitHub Somehow, I could not push my updated project on GitHub because "Push master to origin/master was rejected by remote". I am in the process of fixing this issue.

What's next for Recycling Tracker

I would like to have some user tests and receive feedback to improve features of this app.

Built With

Share this project:

Updates