perf(overlap): drop dead TP/FN/FP/TN pixel-map computation (~3x)#93
Closed
timtreis wants to merge 1 commit into
Closed
perf(overlap): drop dead TP/FN/FP/TN pixel-map computation (~3x)#93timtreis wants to merge 1 commit into
timtreis wants to merge 1 commit into
Conversation
measureobjectoverlap computed TP/FN/FP/TN pixel maps (via the subscripts() set-intersection + maskimg() per-pixel-assignment helpers) and then never used them: the function returns `results`, which only holds the overlap measurements and the optional EMD — the four maps are assigned and discarded, never read, never added to results, never returned. They are a verbatim carry-over from CellProfiler's MeasureObjectOverlap, which renders TP/FP/FN/TN as overlay images; cp_measure returns a feature dict, so they have no consumer here. The `set(zip(...)) & set(zip(...))` intersection plus the per-pixel Python assignment loop was ~2/3 of the function's runtime. Deleting the dead block and the now-unused subscripts()/maskimg() helpers is output-identical and ~3x faster end-to-end (1024px/64 objects: 3140ms -> 1060ms, 2.96x; output verified allclose, 19 multimask tests pass). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Benchmark —
|
| size \ objects | 16 | 64 | 256 |
|---|---|---|---|
| 256 | 1.10× | 1.00× | 1.01× |
| 512 | 1.00× | 0.99× | 1.00× |
| 1024 | 1.01× | 1.00× | 1.00× |
zernike
| size \ objects | 16 | 64 | 256 |
|---|---|---|---|
| 256 | 0.94× | 1.09× | 1.04× |
| 512 | 1.01× | 1.01× | 1.02× |
| 1024 | 1.01× | 0.99× | 1.03× |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
measureobjectoverlapcomputed TP/FN/FP/TN pixel maps (via thesubscripts()set-intersection +maskimg()per-pixel-assignment helpers) and then never used them — the function returnsresults, which only holds the overlap measurements and the optional EMD. The four maps are assigned and discarded: never read, never added toresults, never returned.They're a verbatim carry-over from CellProfiler's MeasureObjectOverlap (which renders TP/FP/FN/TN as overlay images); cp_measure returns a feature dict, so they have no consumer here. The
set(zip(...)) & set(zip(...))intersection + per-pixel Python loop was ~2/3 of the function's runtime.Deleting the dead block and the now-unused
subscripts()/maskimg()helpers is output-identical and ~3× faster end-to-end (1024px/64 objects: 3140ms → 1060ms, 2.96×; verifiedallclose, all 19test_multimask.pytests pass).Net: −32 lines, no behavior change.