GetTrash

Inspiration

GetTrash is inspired by the severe hazards contaminated in our environment. In fact, many animals, such as stray cats and dogs who are seeking for nourishment and get harmed by wastes scattered on the ground. Hence, we are inspired to find a solution to save lots of resources and protect both people, animals, and the environment. Sorting your waste well helps to ensure that hazardous substances are disposed of safely. By separating different materials, most of them can also be recycled into new products.

What it does

GetTrash is a web app that contains two features, live-toggle image scanning and uploaded pictures. In live-toggle image feature, the user is able to connect to their webcam or phone and scan a real-life object with ease. They can also upload an image at their convenience to get a image detection caption underneath the image.

How we built it

GetTrash utilizes Python, TensorFlow, and Streamlit, as well as Google Generative AI to sort both live photo scanning and uploaded image captions.

Challenges we ran into

It took very long for us to train the models under a variety of classifications. Additionally, the model we built was quite large (100mb+) and thus ran into difficulties processing it and sharing it through git and other team members. Additionally, with time constraints, it was hard to format output from the model.

Accomplishments that we're proud of

We are able to get a prototype working for the live-toggle through our video demo and try out our streamlit downloaded picture caption in the try it out link!

What we learned

We learned Streamlit, Tensorflow, and working with Generative AI

What's next for GetTrash

We hope to work on our web app more and improve our accuracy!

Built With

• css3
• flask
• genai
• html5
• javascript
• python
• streamlit
• tensorflow