EcoVision

Inspiration

We wanted to empower individuals to make sustainable choices by providing an easy way to identify recyclable, reusable, or salvageable items around them. Inspired by the growing need to reduce waste, our project offers a user-friendly tool that can help everyone contribute to a greener planet.

What it does

EcoVISION allows users to upload an image or video of items to identify which can be recycled, reused, or salvaged. For individual users, it provides actionable tips on repurposing items to promote sustainability at home. Additionally, EcoVision’s video processing feature can analyze footage from conveyor belts in waste management facilities, automatically detecting recyclable and reusable items on a larger scale. This capability could streamline sorting processes and boost recycling efficiency in industrial waste management, making EcoVISION adaptable from personal use to large-scale environmental solutions.

How we built it

We used a combination of computer vision and machine learning models to analyze images and videos. Our backend, built-in Flask, processes and extracts frames from videos, while the OpenAI API identifies items and categorizes them based on how they can be repurposed. We then present the data in a web-based interface using React.

Challenges we ran into

The biggest hurdle we faced was finding a model we could use to identify items in a given image. Many of the popular models struggled to identify items as they were trained for larger, more common items. To overcome the challenge of identifying smaller, less common items in images, we incorporated OpenCV for detection. OpenCV allowed us to preprocess the images and detect objects effectively by applying custom parameters and filters that suited our specific needs, such as recognizing smaller items in varied conditions. This approach enabled us to improve detection accuracy beyond what popular models alone could achieve.

Accomplishments that we're proud of

We were thrown into an unfamiliar tech stack, specifically using Flask for the backend. We had to learn this new technology under a time crunch to develop our backend and have it communicate efficiently with our React frontend. Our communication and collaborative efforts let us quickly adapt to this unfamiliar technology with minimal effect on overall productivity.

What we learned

Throughout this project, we gained experience in integrating computer vision and machine learning for environmental impact. We learned to adapt to new technologies, specifically using Flask for backend development. Additionally, we worked together to address challenges related to item detection and real-time video analysis. This experience has shown us the potential of technology in creating practical sustainability tools and has motivated us to further explore the intersection of tech and environmental solutions.

What's next for EcoVISION

For EcoVISION, we envision an impactful future with several strategic expansions to drive sustainability in urban environments.

Expand to Waste Management Facilities: One of our primary goals is to implement this solution at waste management facilities. By integrating our technology, these facilities can automatically filter recyclable items from general waste, significantly increasing the recycling rate. This not only reduces the amount of waste going to landfills but also helps cities manage their waste in a more sustainable and eco-friendly manner. Such large-scale application could lead to cleaner, greener cities with less environmental impact.

Better Machine Learning Models: To improve detection accuracy, we plan to focus on improving our machine learning models. Currently, our model can identify many recyclable items, but with further development, it can become even more precise and versatile.

Partnership with Local Recycling Centers: Collaborating with local recycling centers is another critical step. Through these partnerships, we can collect valuable real-world data to refine our model continuously. Recycling centers can provide insights into specific waste streams and help test our system in practical settings, ensuring it meets industry needs. These partnerships will also allow us to create a network for data sharing and model improvement, accelerating our progress and making EcoVISION a more effective solution.