-
-
original test file in demo
Inspiration
We have a hard time understanding how a task as simple as discarding the used bits of product into a waste bin is consistently left incomplete. This waste accumulates and ruins communities, natural habitats, and in some cases cause death and other health-related issues. We decided there must be a way we can use technology to help combat this issue.
What it does
It is an image recognition software that recognizes when an area is littered and what type of litter is in the area.
How we built it
We used openCV's masking functions to filter our pictures. We relied on the fact that trash typically has very bright colors and distinct shapes and sizes to mark out in the give photo. Once our masks filtered and marked our photos, we would crop the parts of the image of just the trash and then we would store these images into a list. After, we took the list of images and sent them to a Tensorflow model we trained on Google's CoLab to recognize what kind of trash the machine is seeing.
Challenges we ran into
None of us had much hardware experience. We did try to connect our software to an arduino but due to limited supplies, knowledge, and energy, we decided to scratch the arduino and continue working on creating a high quality proof of concept.
Accomplishments that we're proud of
We're proud of training a semi- accurate model for trash recognition in such a short amount of time.
What we learned
We learned how to use tensorflow on a cloud, how much quicker running tensorflow on the cloud at a GPU speed is than running on CPU speed, we realized hardware is difficult, and we learned how to take a project one step at a time.
What's next for LitCam
We would like to get hardware involved and include geolocation functionality in our software. We'd also like to connect a backend that can contact cleaning teams for when trash is located. Ultimately, the goal of the software is to detect litter, locate it, recognize the type, and quantify it so that users can get to litter quicker and clean it more efficiently. We'd also like to connect LitCam to social media and find a way to live stream it as a means of creating social awareness for the amount of litter contaminating the planet every day.
Log in or sign up for Devpost to join the conversation.