Fix batching with oriented images and support larger image datasets #178
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
PaulHax
changed the title
- Use generator when computing embeddings. - Ajust image cache size based on image size and available memory.
PaulHax
changed the title
Don't cache the full transformed image in embedding app. Was running out of memory.
alesgenova
reviewed
Feb 7, 2025
| return file_path.split(".")[-1].upper() in COMPATIBLE_FORMATS | ||
|
|
||
|
|
||
| def ensure_browser_compatable_format(image: Image.Image): |
There was a problem hiding this comment.
typo "compatable" -> "compatible"
There was a problem hiding this comment.
|
@alesgenova Should I merge this down, then trigger a release? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The shitspoter dataset has ~12 megapixel images.
Closes #92 <- Was happning when the transformers.pipeline was rotating shitspoter images in preprocessing based on their medatdata. That broke the "all images in batch must be same width and height" rule, causing the error.