Trashnet Neural Network

• 

  We tried to add our neural network to our website but this proved too difficult to accomplish because PyTorch cannot be modified into JS.

Inspiration

According to the Environmental Protection Agency (EPA), there were about 60-75 million tons of products that could have been recycled but were thrown in the trash (in the USA). This is an astronomically large number and something that we thought we could fix.

What it does

Our neural network takes in an image and classifies it based on whether it is recycling or trash. This was supposed to be added to our website which would allow users to take an image of an object and see whether it should be put in the trash or recycled.

How we built it

Using python and pytorch, along with the help of github pages and outside papers on neural networks, we managed to create a basic neural network structure that uses convolutions to scan images and give a final answer of what an image might be.

Challenges we ran into

One challenge that we ran into was that we were unable to add our neural network to our website. This was because PyTorch cannot be changed into Javascript and we did not have enough time to change our neural network.

Accomplishments that we're proud of

We are proud of being able to complete a neural network within the time frame, even though we were not able to add it to our website.

What we learned

We learned that it is important not to do too much because we ran out of time to do everything we wanted. We also learned that it is important to do research and make sure that what you are planning to do will work. If we chose to build the neural network using Tensorflow, we would have been able to implement it in our website.

What's next for Trashnet Neural Network

We plan to switch it to Tensorflow and train it using a larger dataset. This would improve the accuracy of the neural network and allow us to add it to our website.

Built With

• python
• pytorch
• replit
• trashnet