Public welfare AI

Public welfare AI 项目故事

Inspiration

My inspiration for Public welfare AI comes from UNEP’s environmental reports and Google’s AI for Social Good initiative. Shocking data on plastic pollution and deforestation awakened my sense of responsibility, while Google’s practice showed me AI’s potential in environmental protection—using technology to simplify environmental actions, which drove me to launch the project.

What it does

Based on Google’s initiative, Public welfare AI is a simple AI tool for environmental protection, with three core functions: 85%+ accurate pollution source identification via CNN, intuitive environmental data visualization, and simple local environmental change prediction using linear regression models, helping the public and grass-roots organizations participate in environmental protection easily.

How we built it

We built it in 3 stages: first, clarified needs by researching grass-roots environmental organizations; second, selected lightweight technologies (Python, CNN, linear regression, Streamlit) and collected public environmental data; third, trained and optimized models to ensure practicality and simplicity.

Challenges we ran into

We faced two main challenges: insufficient high-quality pollution images for model training, and balancing model accuracy with tool lightness. Simple algorithms affected accuracy, while complex ones increased operation difficulty.

Accomplishments that we're proud of

We are proud that the tool achieved over 85% pollution identification accuracy and a user-friendly interface. It has been used by several grass-roots organizations, truly lowering the threshold for environmental protection.

What we learned

We learned to combine AI technology with real environmental needs, master lightweight algorithm application, and solve problems flexibly. More importantly, we realized that technological public welfare lies in practicality, not complexity.

What's next for Public welfare AI

Next, we will improve model accuracy, expand pollution identification types, cooperate with more public welfare organizations, and optimize the interface to make Public welfare AI more accessible and better serve environmental protection.

Built With

• python