Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Can we go to support displaying the names of labels in Run Prediction UI? #10

Open
chrishkchris opened this issue May 7, 2020 · 2 comments
Assignees

Comments

@chrishkchris
Copy link
Contributor

Currently the Run Prediction UI is:

77987199-c995ca80-734b-11ea-8f0f-c5f65c7bc135

In the MC dropout result, we don't know the name of the prediction label and what are the names of the classes 0, 1, 2, …, 9
The predicted class should be the class with the highest mean in the MC dropout distribution.

Therefore, can we do the following adjustments in the current Run Prediction UI

  1. Display the name of the prediction result (e.g. X-ray: Positive or Negative)
  2. Display the names of the classes in MC dropout (in additional to the class number 0 - 9)
  3. Display also the original image without explanation

To understand what is the name of a label means, we can look at the following figure

77987545-ad465d80-734c-11ea-8011-1f8ada4f11c1

To clarify, we don't need to change our UI design, we just need to 1. add the name of the labels, 2. the prediction result and 3. the original image without explanation on it.

@nudles
Copy link
Contributor

nudles commented May 7, 2020

The solution is to move this prediction page outside of singa-auto system, and update the UI for image classification tasks.
#8

@chrishkchris
Copy link
Contributor Author

@arielclj

The existing inference job of image classification is attached to the existing UI, can it be migrated into the new web folder following the rule in #8

You may take the following step:

  1. migrate the existing image classification application first
  2. deploy and demo it to us
  3. modify the web apps to address this issue

You can open a new PR for that

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants