The Plant Doctor

Inspiration

We were inspired by

What it does

Our project features a web page called "The Plant Doctor". On the site, people can upload images of sick plants and the site will identify the disease.

How we built it

The image identification is done thought machine learning. We trained a VGG16 ML model on an image dataset from kaggle. The model was trained on Google Colab and the weights were transferred later. The front-end was made with vanilla HTML and CSS, and Flask is being used to link the two together.

Challenges we ran into

We had plenty of challenges while making this. We had issues uploading the image dataset (from Kaggle) to our code. We switched to Google Colab to train our model. We also had trouble connecting our front-end with our flask server. However, after some advice from the organizers, we were able to get them connected.

Accomplishments that we're proud of

We are proud that we were able to train our model Most importantly, we are proud of our ability to function this well with so little sleep.

What we learned

We learned a lot of things at this Hackathon. We gained experience in machine learning by training our model and implementing it. We also learned how integrate the front-end and back-end.

What's next for Plant Doctor

We plan to improve our UI to be more user-friendly, and to style our results page. We also plan to improve the model and make transfer it to its own webpage.