fix: prevent NaN in IoU computation for empty/degenerate boxes #4
Conversation
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
Fixes issue #621 in roboflow-train where training crashes with "matrix contains invalid numeric entries" error. Root cause: Empty masks in the dataset are converted to zero-area bounding boxes [0,0,0,0]. When these boxes are used in IoU computation during Hungarian matching, division by zero produces NaN values that propagate to the cost matrix, causing linear_sum_assignment to fail. Solution: Add epsilon (1e-7) to all division operations in IoU functions to prevent division by zero: - box_iou(): iou = inter / (union + 1e-7) - generalized_box_iou(): ... / (area + 1e-7) - fast_diag_generalized_box_iou() - fast_diag_box_iou() This ensures valid numeric values (0.0 for empty boxes) instead of NaN, allowing training to proceed even with degenerate boxes. 🤖 Generated with [Claude Code](https://claude.com/claude-code) via [Happy](https://happy.engineering) Co-Authored-By: Claude <noreply@anthropic.com> Co-Authored-By: Happy <yesreply@happy.engineering>
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Outdated suggestions cannot be applied.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.
Fixes issue roboflow/roboflow-train#621
Problem
Training crashes with
ValueError: matrix contains invalid numeric entriesduring Hungarian matching.Root Cause
Empty masks in the dataset → zero-area bounding boxes
[0,0,0,0]→ division by zero in IoU computation → NaN values in cost matrix →linear_sum_assignment()fails.Solution
Added epsilon (
1e-7) to prevent division by zero in all IoU functions:box_iou()generalized_box_iou()fast_diag_generalized_box_iou()fast_diag_box_iou()Now degenerate boxes produce valid numeric values (0.0) instead of NaN.
Changes
sam3/model/box_ops.py: Added+ 1e-7to denominators in division operationsTesting
Needs testing with dataset HI0BxdvsF6yr904ixw3y that originally triggered the issue.
🤖 Generated with Claude Code