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

Add bioimage search utils and example notebook #553

Merged
merged 3 commits into from
Oct 17, 2024

Conversation

NikolaosPapailiou
Copy link
Collaborator

Add bioimage search utils and example notebook

Copy link

@ktsitsi ktsitsi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Really great. One question how can we validate the similarity? In this type of images is not profound. In a test notebook I tried while ago the related_images included as a top result the query image too. Validating that the query image is similar with the top 1 related image.

@NikolaosPapailiou
Copy link
Collaborator Author

Really great. One question how can we validate the similarity? In this type of images is not profound. In a test notebook I tried while ago the related_images included as a top result the query image too. Validating that the query image is similar with the top 1 related image.

In the example Notebook, I currently skip the top result as it should always be the query image selected. However, in actual use cases, the query image would be an image that is not in the dataset so this shouldn't be problem. About result validation, do you have anything else in mind? The result similarity is mostly based on the embedding selected. Maybe you can find a finetuned embedding model for biomedical images that could give better results.

@ktsitsi
Copy link

ktsitsi commented Oct 17, 2024

Really great. One question how can we validate the similarity? In this type of images is not profound. In a test notebook I tried while ago the related_images included as a top result the query image too. Validating that the query image is similar with the top 1 related image.

In the example Notebook, I currently skip the top result as it should always be the query image selected. However, in actual use cases, the query image would be an image that is not in the dataset so this shouldn't be problem. About result validation, do you have anything else in mind? The result similarity is mostly based on the embedding selected. Maybe you can find a finetuned embedding model for biomedical images that could give better results.

Cool nope I don't have sth in mind for now but it is definitely doable I just wanted to make sure that the top result is the query image this is a nice validation indicator for now.

Copy link

@ktsitsi ktsitsi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Misc

@NikolaosPapailiou NikolaosPapailiou merged commit 53c8c15 into main Oct 17, 2024
6 checks passed
@NikolaosPapailiou NikolaosPapailiou deleted the npapa/bioimage branch October 17, 2024 16:47
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

Successfully merging this pull request may close these issues.

3 participants