Skip to content

README.md

Latest commit

 

History

History
45 lines (28 loc) · 1.49 KB

README.md

File metadata and controls

45 lines (28 loc) · 1.49 KB

webcamKnn

Base project for using the webcam and running the results through mobileNet, and the results of that through a KNN to allow for arbitrary classes to be discovered based off a small example set.

How to use

Clone the repo, and run basicFlaskApp.py after this go to localhost:5000 and you should get a prompt for using your webcam.

For extra customization you can add some path arguments example

http://localhost:5000/?numOfClassOne=2&numOfClassTwo=5&numOfClassThree=3

http://localhost:5000/?numOfClassTwo=2

Requirements

Requires flask

pip installing that should pull all required dependencies

Either python 2 or 3 should work for this project

Future work

This is the base for the class project there/not there, and light/dark sensor.

TODO

  • Fix the load button
  • Fix the save button
    • both of these buttons need to have all three classifiers so they can properly read and write the data from/to file

Done

  • Add another KNN with a different set of classes
  • Add a third KNN with the input being the two previous KNN's
    • This will be the base for the overall solution
  • Migrate application to a flask service
    • This will allow us to serve the application on something like heroku easily in the future
  • Create a better way to index the labels and text
    • Potentially use a loop over some variables

Base of this project

This is heavily based off of the teachable machine Google tool, however simplified for easy addition to personal projects.