-
-
User interface for website
-
Image based classification
-
Live classification
-
chat interface
-
End goal
Inspiration
Our inspiration for RecyliVision came from the growing need to tackle the global waste crisis. With millions of tons of waste being improperly disposed of, we wanted to create an AI-driven solution that not only identifies waste categories but also educates people on how to properly decompose it and understand its lifecycle. The idea of combining image classification and a chatbot to aid in waste management felt like a powerful step towards a cleaner planet.
What it does
It classifies waste into three categories: recyclable, non-recyclable, and organic. By analyzing images of waste using advanced convolutional neural networks (CNN), the system provides accurate sorting. Additionally, it features a chatbot that assists users by offering guidance on how to decompose different types of waste and explaining their lifecycle, making the waste management process more accessible and educational.
How we built it
We built it using a large waste dataset and leveraged Convolutional Neural Networks (CNN) for image classification. We also integrated OpenCV for image preprocessing and real-time analysis. The backend is powered by machine learning algorithms trained to distinguish waste types with high accuracy. For the chatbot, we implemented a natural language processing system to answer queries related to waste decomposition and environmental impact.
Challenges we ran into
One of the primary challenges we faced was data collection. Building a comprehensive and diverse dataset for waste classification was time-consuming and required significant effort. We also encountered difficulties in optimizing the model's performance for different lighting conditions and varying waste materials. Another challenge was integrating the chatbot with accurate and reliable waste lifecycle information, ensuring it provides useful advice.
Accomplishments that we're proud of
We are proud of developing a functional prototype that accurately classifies waste and helps users understand how to dispose of waste properly. The integration of both image classification and chatbot features is a major milestone for us. Additionally, overcoming the challenge of building a robust dataset and optimizing our CNN model for various conditions was a rewarding accomplishment.
What we learned
Through this project, we learned the importance of dataset diversity in training AI models. We also gained a deeper understanding of image processing techniques using OpenCV and CNNs. Moreover, working on the chatbot taught us the value of providing educational resources along with technological solutions, combining AI with real-world applications for environmental impact.
What's next
The next step for it is to implement a drone system equipped with cameras capable of analyzing the location of waste in real time. The goal is to autonomously identify waste in outdoor environments and classify it, making the solution even more scalable and efficient. We're also looking into further improving the chatbot's capabilities and adding more detailed lifecycle information for a wider variety of waste types.
Log in or sign up for Devpost to join the conversation.