1
$\begingroup$

I am working on a problem with images where we are monitoring development of spot in certain region of image. We are able to classify spot present(NOK) or not present(OK) successfully if initially image doesn't have spot in that region and developed over time by using CNN with ReLU activation. But our challenge is in finding solution for images which already have spot initially (which is OK in this case) but it developed more overtime (NOK). If we feed 2nd scenario of data in our trained model(CNN with ReLU), it starts classifying as spot present, which is desired outcome for 1st scenario but not the desired outcome for 2nd scenario.

Note : images are collected for same component from different machines, where some of the component's region is clean with no initial spot observed and some of the component have some mild spot, which is OK in our case.

My challenge is:

  1. how to create a ML model which can understand baseline of the specific region and classify growth of spot observed in that specific region of image for both scenario stated above ?
$\endgroup$

1 Answer 1

0
$\begingroup$

I found my answer in data pre - processing steps. Finding the growth of spot or stain was the major challenge for the cases where we have already stain. Calculating delta with respect to initial image can clearly show the evidence of growth in stain. So we decided to train the model on the pre-processed data by get relative image and use same for train and test purpose.

$\endgroup$
1
  • $\begingroup$ Your answer could be improved with additional supporting information. Please edit to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers in the help center. $\endgroup$
    – Community Bot
    Commented Nov 26, 2021 at 20:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.