Run configuration

Steps with full coverage
2/9
Steps with partial coverage
6/9
Steps with no coverage
1/9
Top raw false positive
18986
126 detections

Per-step summary

StepStatusExpected partsHit framesWhat actually hitMissing
1
step_001_correct_0.mov
partial30029, 30235/730029 (5/7)3023
2
step_002_correct_1.mov; step_002_corrected_2.mov
partial35480, 368412/1436841 (2/14)35480
3
step_003_correct_3.mov
none204820/7none20482
4
step_004_correct_4.mov
partial3710, 997813/799781 (3/7)3710
5
step_005_correct_5.mov
full3023, 36664/73023 (2/7), 3666 (4/7)
6
step_006_correct_6.mov
full368403/736840 (3/7)
7
step_007_correct_7.mov
partial3022, 37107/73022 (7/7)3710
8
step_008_correct_8.mov
partial28802, 30241/728802 (1/7)3024
9
step_009_correct_9.mov
partial4032a, 997801/74032a (1/7)99780

Ranked actionable error analysis

Highest-impact misses to fix first

  1. 3710 — hit rate 0/14 (0%); blocking steps 4, 7; recoverable missed step-frames: 14.
  2. 3023 — hit rate 2/14 (14%); blocking steps 1, 5; recoverable missed step-frames: 12.
  3. 35480 — hit rate 0/14 (0%); blocking steps 2; recoverable missed step-frames: 14.
  4. 36841 — hit rate 2/14 (14%); blocking steps 2; recoverable missed step-frames: 12.
  5. 20482 — hit rate 0/7 (0%); blocking steps 3; recoverable missed step-frames: 7.
  6. 3024 — hit rate 0/7 (0%); blocking steps 8; recoverable missed step-frames: 7.
  7. 99780 — hit rate 0/7 (0%); blocking steps 9; recoverable missed step-frames: 7.
  8. 28802 — hit rate 1/7 (14%); blocking steps 8; recoverable missed step-frames: 6.

Read: rank is by how many steps a class blocks first, then by missed frame count. That makes 3710 and 3023 the highest-leverage fixes because they hurt multiple steps.

Raw false positives (unconstrained detector behavior)

Design IDCountShare of top-10 FP mass
1898612631.6%
30229323.3%
68064711.8%
30029348.5%
36840266.5%
3005184.5%
28974174.3%
4032a133.3%
20482133.3%
4006123.0%

Top-10 false positives account for 399 raw detections. The biggest offenders are driving review noise and should be suppressed or rebalanced during retraining.

Bottom line

Actionable next moves