PetriAI

nspiration

We built PetriAI because counting bacterial colonies by hand is slow and easy to mess up. Students, teachers, and small labs may not have expensive lab equipment, so we wanted to make petri dish analysis faster and easier with AI.

Rubric: This is unique because it turns a normal petri dish photo into useful lab results.

What it does

PetriAI lets users upload a petri dish image and gives them:

colony count growth coverage growth level contamination review flag automated lab report

Instead of only showing numbers, it explains what the results mean.

Rubric: This matches the prompt because it solves a real-world lab problem with an AI startup MVP.

How we built it

We built PetriAI as a web app with a frontend, Flask backend, image upload, and SQLite history.

The app uses computer vision to detect the dish, find colony-like spots, calculate coverage, and generate a report. It uses OpenCV methods like contour detection, adaptive thresholding, and dish boundary detection.

Rubric: This shows we built a working technical product, not just an idea.

Challenges we ran into

The hardest part was making the app work with real petri dish images. Images can have glare, shadows, blurry spots, handwriting, or uneven lighting.

Another challenge was making the results easy to understand, so we added a lab report instead of only showing numbers.

Accomplishments that we're proud of

We are proud that PetriAI is a complete MVP. It has image upload, analysis results, contamination flagging, report generation, and saved history.

We are also proud that the idea has real users, like biology students, teachers, college labs, and science fair students.

Rubric: It is scalable because it can start in schools and later expand to universities, food safety training, and small labs.

What we learned

We learned how computer vision can turn an image into useful scientific data. We also learned that a good AI project needs more than AI; it needs a clear problem, real users, and results that people can understand.

What's next for PetriAI

Next, we would improve accuracy by testing on more petri dish images. We would also add editable annotations, comparison mode, better AI-generated reports, and online deployment.

In the future, PetriAI could use a free plan with limited scans and paid plans for schools, labs, and organizations.

Final rubric connection: PetriAI is unique, scalable, and directly fits the prompt because it is an AI startup MVP that solves a real lab problem.

Built With

• api
• kaggle
• machine-learning
• sqlite