Skip to content

Conversation

@bkowshik
Copy link
Contributor

While cleaning up datasets, we want to look at just the images and their labels. The current rs compare requires at-least one masks directory and this PR makes the masks directory optional, one ore more to be precise! Now, we could use this tool to generate images like below:

95309

@bkowshik bkowshik requested a review from daniel-j-h July 16, 2018 08:49
Copy link
Collaborator

@daniel-j-h daniel-j-h left a comment

Choose a reason for hiding this comment

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

Should this tool handle arbitrary slippy map directories? As in: if you pass it slippy map dirs for images and masks it will show you that, if you pass it slippy map dirs with images, masks, probs, and predicted masks, it will also show you that.

@bkowshik
Copy link
Contributor Author

👍

Things I have tested rs compare for are:

  • Slippy map dirs for images and masks (this branch)
  • Slippy map dirs with images, masks and predicted masks

Something that I have not done with this tool is the following but, I don't see any blockers for something like this not to work.

  • Slippy map dirs with images, masks, probs, and predicted masks

One additional functionality that rs compare tool handles is filtering by min/max percentages of masks not background. The use-case here is to filter out images that have predictions above or below a threshold. Breaking that functionality out of this tool would allow us to create a generic rs compare that given two or more slippy map directories of images, combines them into one image.

Did I understand you correctly?

@daniel-j-h
Copy link
Collaborator

What I meant was

  • removing the threshold check since it's orthogonal to this tool's core responsibility
  • taking a list of paths to slippy map directories from the command line
  • walking all slippy maps in parallel, creating one image with all slippy map tiles next to each other

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.

2 participants