Skip to content

A webapp that uses machine learning via The BeeImage Dataset to generate an improving model for classifying a bee's health

License

Notifications You must be signed in to change notification settings

liv-yaa/ML_BeeHealth

Repository files navigation

Project title

"The Bee Machine"

Motivation

The Bee Machine is a tool that can help researchers by "crowd sourcing" health data about honeybee populations from around the world. A user anywhere can submit a photo of a bee and tag its location. Once the image is processed by the Bee Machine's machine learning model, a prediction of that honeybee's health will be returned. If the prediction is not accurate, the model will learn from its mistakes as well as its successes. Over time, statistics show that the model will return fewer false positive and false negative predictions and more true positive and true negative predictions. This way, any user who wants to contribute to our database can help track honeybee health trends.

Created with an a eye toward preserving our honeybee populations, which are intricately intertwined with the food and agricultural systems we have relied on for millenia. Preserving niche species and promoting biodiversity will be a challenge in the coming years, but with more research by honeybee immunologists and population studies, we can start to gather better metrics about honeybee health.

Tech and frameworks used:

  • Python
  • Flask
  • Jinja
  • CSS
  • HTML
  • SQLAlchemy
  • PostgreSQL Database
  • Bootstrap
  • Delployed & 1.1E+4 image database hosted on AWS

How to use?

Upload a photo. Specify "healthy" for healthy bees. All other enties will be tagged "unhealthy".

The computer vision algorithm will use your submission to become better at bee health classifications.

Contribute

You can contribute to The Bee Machine by uploading a bee photo. The more photos are submitted, the more powerful the health sorting algorithm will become. General guidelines for photo submissions would be to make sure the image is quality and it is a bee. Otherwise, the model should screen for any non-bee images too.

Credits

Built with

Sage Advice



About

A webapp that uses machine learning via The BeeImage Dataset to generate an improving model for classifying a bee's health

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published