Validation error analysis

12 selected cases, six shared categories. Confidence ≥ 0.25; matching IoU ≥ 0.5. Ground truth (lime), matched detections (green), false alarms (orange), ignored detections (gray). Click images for full-size views. Counts use category-aware COCO matching with crowd ignores. Ignore-region outlines are available in the interactive website, not these static panels. Cases show selected extremes, not average performance.

01 · Original site frame (fixed before prediction inspection)

0000026_00500_d_0000025.jpg

Matched objects 1 → 11 (+10); false alarms 0 → 2; fine-tuned misses 4/15. 10 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 1baseline for case 1fine_tuned for case 1

02 · Original site frame (fixed before prediction inspection)

0000242_00627_d_0000003.jpg

Matched objects 2 → 9 (+7); false alarms 1 → 6; fine-tuned misses 30/39. 7 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 2baseline for case 2fine_tuned for case 2

03 · Original site frame (fixed before prediction inspection)

0000194_00800_d_0000123.jpg

Matched objects 2 → 6 (+4); false alarms 2 → 5; fine-tuned misses 36/42. 4 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 3baseline for case 3fine_tuned for case 3

04 · Largest increase in matched objects

0000129_02411_d_0000138.jpg

Matched objects 0 → 77 (+77); false alarms 0 → 39; fine-tuned misses 89/166. 77 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 4baseline for case 4fine_tuned for case 4

05 · Largest increase in matched objects

0000271_06001_d_0000402.jpg

Matched objects 9 → 61 (+52); false alarms 2 → 32; fine-tuned misses 90/151. 52 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 5baseline for case 5fine_tuned for case 5

06 · Largest increase in matched objects

0000295_02000_d_0000031.jpg

Matched objects 3 → 54 (+51); false alarms 1 → 19; fine-tuned misses 252/306. 51 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 6baseline for case 6fine_tuned for case 6

07 · Largest decrease in matched objects

0000103_01734_d_0000028.jpg

Matched objects 4 → 0 (-4); false alarms 1 → 1; fine-tuned misses 4/4. 0 newly matched objects and 4 lost matches at this display threshold.

ground_truth for case 7baseline for case 7fine_tuned for case 7

08 · Largest decrease in matched objects

0000023_01233_d_0000011.jpg

Matched objects 5 → 2 (-3); false alarms 2 → 4; fine-tuned misses 12/14. 1 newly matched objects and 4 lost matches at this display threshold.

ground_truth for case 8baseline for case 8fine_tuned for case 8

09 · Largest decrease in matched objects

0000024_00500_d_0000013.jpg

Matched objects 4 → 2 (-2); false alarms 1 → 2; fine-tuned misses 8/10. 0 newly matched objects and 2 lost matches at this display threshold.

ground_truth for case 9baseline for case 9fine_tuned for case 9

10 · Most remaining missed objects

0000291_03801_d_0000887.jpg

Matched objects 1 → 10 (+9); false alarms 1 → 10; fine-tuned misses 169/179. 9 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 10baseline for case 10fine_tuned for case 10

11 · Most remaining missed objects

0000086_01954_d_0000005.jpg

Matched objects 0 → 17 (+17); false alarms 0 → 30; fine-tuned misses 156/173. 17 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 11baseline for case 11fine_tuned for case 11

12 · Most remaining missed objects

0000001_08414_d_0000013.jpg

Matched objects 1 → 16 (+15); false alarms 0 → 10; fine-tuned misses 138/154. 15 newly matched objects and 0 lost matches at this display threshold.

ground_truth for case 12baseline for case 12fine_tuned for case 12