Fix smart_resize to use actual image dimensions instead of fixed config values#1032
Merged
apsonawane merged 2 commits intomainfrom Mar 5, 2026
Merged
Fix smart_resize to use actual image dimensions instead of fixed config values#1032apsonawane merged 2 commits intomainfrom
apsonawane merged 2 commits intomainfrom
Conversation
…nfig dimensions Pass actual image (h, w) to smart_resize() instead of fixed config (height_, width_). This preserves the aspect ratio and matches HuggingFace smart_resize behavior, significantly improving accuracy on vision-language benchmarks.
apsonawane
approved these changes
Mar 5, 2026
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem:
In Resize::Compute(), smart_resize was called with the fixed config dimensions (height_, width_) instead of the actual input image dimensions (h, w). This caused all images to be resized to the same target size regardless of their original aspect ratio, producing incorrect preprocessing results that diverged from HuggingFace's smart_resize behavior.
Fix:
Pass the original image dimensions to smart_resize() so the target size is computed based on the actual aspect ratio, snapping to the nearest patch grid boundary while respecting min_pixels / max_pixels constraints. This matches the HuggingFace Qwen2VLImageProcessor behavior exactly.