Our project has three main parts. First, we designed our own accessible and convenient flying drone from accessible hardware, using brushless motors, ESCs, and a GPS to help detect waste in areas such as beaches, parks and other clean spaces.
Second, we used a pre-trained Edge Impulse model with the Arduino Uno Q to recognize waste. The model has a F1 Score with accuracy above 90%. When waste is detected, the system can trigger a live web feed that the user can view directly in a browser. It will also reproduce a confidence level to let the user know whether there is garbage within the image classification surveillance.
Finally, the system sends an email notification when waste is detected. The email informs the user that waste has been found and provides the relevant information needed to locate and clean it up, making the process faster and more automated. The relevant information includes timestamp, date and an automated message that the user can access from their gmail.
Log in or sign up for Devpost to join the conversation.