Our team made an Android app allowing users to categorize the garabge and learn if it is recyclable. Users can also learn how to recycle through the hyperlink.
We utilized Android Studio using Kotlin, and trained Machine Learning Model (Tensorflow Lite) through TeachableMachine.withgoogle.com with around 2,000 datasets.
There are some limitations especially for detecting glass and metals. Thus, we should improve our detecting model with more/better images.
Inspiration
Environment is the most important part of our life and we humans are producing so much trash that mother earth is now polluted so much that we need some good innovation to fight this and solve the problem.That's why we come with categorical trash,an android app.
What it does
It basically a very simple application which takes picture from the users phone camera and after that it differentiate the type of the junk in various categories like metal,glass,biodegradable,plastic and cardboard.You got a link after that which give you the good reference what to do with the trash.
How we built it
We basically search for some similar opensource code to reuse in our application,so we found the android app from the github and we train a model from the google colab and teachablemachine.Now we collected data from the trashnet.After that we get the tflite model and embedded into our application to run it.Voila it's works awesome.
Challenges we ran into
- first challenge is to get data and we fill that we lack of data.
- Time constraint to train a model.
- Getting ready with sceleton application to make our demo work.
Accomplishments that we're proud of
- We built it within 18 hours and now going to improve it more.
- Our team worked together and created this awesome application this far.
What we learned
- Machine Learning with tflite
- Android app development using Kotlin
- Some basic UI design
What's next for Categorical trash
- Adding more functionality
- Publishing it to the play store
- Getting feedback from users to improve.
Built With
- android-studio
- kotlin
- teachable-machine
- tensoflow-lite
- tensorflow

Log in or sign up for Devpost to join the conversation.