SayNoToTrash

• 

  SayNoToTrash

• 

  Detecting trash from Google Street View

Inspiration

Texas litter survey. This survey is done around every 6 years and we believe we can leverage Google Streetview in order to give preliminary analysis

What it does

Detecting trash metrics with neural networks. It helps finding trash hotspots easier, to lower the climate impact caused by waste

How we built it

Using TensorFlow to detect trash within streetviews images to create trash metrics

Challenges we ran into

We ran into several challenges the first being the use of Streetviews API, where every image requested is capped at a 640x640 resolution. Secondly was the limitations of the data set used, TACO a trash dataset is not targeted at images captured by Streetview but rather closer images from a phone

Accomplishments that we're proud of

Being able to set up image detection using TensorFlows python library and implemented a neural network that can detect trash within an image

What we learned

We learned concepts in machine learning and neural networks as well as managing API's and making efficient calls

What's next for SayNoToTrash

Adding more feature such as pushing to cloud/browser using TensorFlow.js. Computing at optimal times to use more sustainable energy.

Built With

• angular.js
• kaggle
• python
• tensorflow