Banana ripeness status identification

Have you ever been uncomfortable eating a mushy banana? Finding it overripe when peeled? We have the solution, just snap a pic and the model that can identify whether a banana is suitable for eating.

Comment

Inspiration

We looked at one of the biggest popular faucets of AI use. Identification! We believe this interesting aspect of AI is really useful in assisting everyday people on picking out the best quality of food. Also we all don't like mushy bananas.

What it does

This model looks at banana images and gauges the ripeness.

How we built it

For model: We use numpy and Keras of Tensorflow to do data preprocessing, CNN of TensorFlow for the main model, and search around 200 images for each kinds of unripe bananas, ripe bananas, and rotten bananas as training/testing dataset. For website: We use vite with React and Javascript as front end. Python as backend. Flack for API.

Challenges we ran into

Time constraints. Dataset is not easy to fine. CORS have harder standard for browser, this make us use more then 3 hours find the bug but it is just standard problem. And wasting our 3 hours make another website.

Accomplishments that we're proud of

Although the sample size is small, the accuracy of the model's judgment is still high.

What we learned

Mac and window have many more different wait us to find. Small dataset for training AI don't always bad.

What's next for Banana ripeness status identification

To expand upon other food groups!

Built With

Share this project:

Updates