K-Lane evaluation results
Single-frame full run and temporal evaluation snapshot · 2026-08-31 11:28:34 IST
Single-frame evaluated
7,687
of 7,687 · 0 failed
Single-frame final F1
33.12%
surface filter -0.48 pp
Temporal targets processed
1,116 / 1,116
100.0% complete
Valid temporal windows
1,013
103 single-frame fallbacks
Full single-frame evaluation
All labeled frames. K-Lane-tolerant micro metrics compare the range-local baseline with the local-surface-filtered pipeline.
| Pipeline stage | Range-local baseline | Surface-filtered | F1 change | ||||
|---|---|---|---|---|---|---|---|
| Precision | Recall | F1 | Precision | Recall | F1 | ||
| Stage 5 · raw candidates | 33.06% | 28.77% | 30.77% | 36.39% | 28.50% | 31.97% | +1.20 pp |
| Stage 6 · cleaned candidates | 39.54% | 25.38% | 30.92% | 39.98% | 24.75% | 30.58% | -0.34 pp |
| Stage 7 · fitted output | 40.61% | 28.65% | 33.60% | 39.91% | 28.31% | 33.12% | -0.48 pp |
Temporal evaluation · complete
Metrics below include only the 1,116 targets checkpointed when this page was built. The running evaluator has not been stopped or rerun.
| Population and method | Precision | Recall | F1 |
|---|---|---|---|
| All processed · single frame | 33.93% | 22.20% | 26.84% |
| All processed · temporal with fallback | 35.24% | 24.46% | 28.88% |
| Valid windows only · single frame | 34.67% | 22.86% | 27.55% |
| Valid windows only · ten-frame consensus | 36.04% | 25.33% | 29.75% |
End-to-end temporal F1 change on all processed targets: +2.04 pp. Valid-window-only change: +2.20 pp.
Temporal progress and results by sequence
| Sequence | Processed | Valid | Fallback | Single F1 | Temporal F1 | Change |
|---|---|---|---|---|---|---|
| seq_01 | 75 / 75 | 73 | 2 | 27.58% | 28.65% | +1.07 pp |
| seq_02 | 75 / 75 | 70 | 5 | 17.44% | 18.88% | +1.44 pp |
| seq_03 | 75 / 75 | 74 | 1 | 18.53% | 21.05% | +2.53 pp |
| seq_04 | 75 / 75 | 68 | 7 | 16.07% | 16.09% | +0.02 pp |
| seq_05 | 75 / 75 | 70 | 5 | 13.68% | 14.88% | +1.20 pp |
| seq_06 | 66 / 66 | 28 | 38 | 14.90% | 15.60% | +0.70 pp |
| seq_07 | 75 / 75 | 75 | 0 | 26.77% | 26.31% | -0.47 pp |
| seq_08 | 75 / 75 | 62 | 13 | 14.59% | 14.21% | -0.38 pp |
| seq_09 | 75 / 75 | 74 | 1 | 13.29% | 15.07% | +1.78 pp |
| seq_10 | 75 / 75 | 64 | 11 | 20.92% | 22.64% | +1.71 pp |
| seq_11 | 75 / 75 | 65 | 10 | 28.53% | 32.80% | +4.28 pp |
| seq_12 | 75 / 75 | 71 | 4 | 36.88% | 39.57% | +2.69 pp |
| seq_13 | 75 / 75 | 75 | 0 | 44.11% | 50.03% | +5.91 pp |
| seq_14 | 75 / 75 | 71 | 4 | 46.17% | 49.51% | +3.34 pp |
| seq_15 | 75 / 75 | 73 | 2 | 46.39% | 49.19% | +2.80 pp |
Temporal examples · three worst and three best changes
Single 44.80% → temporal 13.43% (-31.37 pp)
Single 46.41% → temporal 16.06% (-30.34 pp)
Single 43.77% → temporal 14.57% (-29.20 pp)
Single 0.00% → temporal 42.68% (+42.68 pp)
Single 25.54% → temporal 60.82% (+35.28 pp)
Single 0.00% → temporal 34.65% (+34.65 pp)