Skip to content

A cross-platform AR Visualizer utilizing a Radial-kernel-trained Support Vector Machine (SVM) model to generate interactive 3D model to understand machine learning concepts

License

Notifications You must be signed in to change notification settings

shah-deep/AR-ML-Fusion

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 

Repository files navigation

First time here? Go to: Steps to Run


This is a tool to Visualize the Machine Learning concept of Support Vector Machines - Radial Basis Function. It trains a model with a dataset containing 2 distinct sets of data points using the SVM Radial kernel. The user can add more points to the data set and change training parameters to re-train the model.

It is a cross-platform web-based tool that can run on a web browser on both Laptops and Mobile Phones. It is an Image Marker-based AR application and uses the Kanji Image marker.

Hosted at: https://shah-deep2.github.io/SVM_AR

Inspired from: SVM Demo

Made with ❤️ at eCampus SJSU

Kanji Image Marker


Kanji Image


Steps to Run

  1. Go to the webpage: https://shah-deep2.github.io/SVM_AR
  2. Grant all the permissions asked including access to the camera. This is required.
  3. Point the camera towards the Kanji Image Marker such that it is fully visible in the camera.
  4. A pre-trained 3D model for SVM-RBF should appear on the screen on top of the Kanji Image Marker.
  5. On a mobile device, use your fingers to rotate and zoom in/out the model. On a laptop/computer, use the mouse cursor to rotate the model.
  6. Click the red/green buttons with a plus sign (+) to add new red/green data points to the model. Click again for every data point you want to add.
  7. Press the grey re-model button (below the red button) to re-train the ML model and re-render the plot.
  8. Click the (i) info button to hide/show the info related to the model.
  9. When model info is visible after pressing the (i) button, you can control the parameters C and σ with the help of a slider.
  10. Moving the sliders to change C or σ automatically re-trains the model.
  11. Play around with the tool by adding different data points and testing different parameters.

Output

3D Model

model

Full Application View on Mobile Phone

full_app