| Milestone | Date | Status |
|---|---|---|
| Preliminary submission deadline | 2026-09-01 23:59:59 | PAST DATE · outcome not recorded |
| Announcement of finalists | 2026-09-03 | PAST DATE · outcome not recorded |
| Finalist submission deadline | 2026-09-15 23:59:59 | PAST DATE · outcome not recorded |
| Experiment | Evaluation scope | Measured result | Decision |
|---|---|---|---|
| MI-DETR1280 E13 · blind replacement | AIdea hidden · same historical pipeline | 0.7689250 vs 0.7701133 · -0.0011883 | REJECTED RECIPE · accepted standalone unchanged |
| MI-DETR1280 E13 · recalibrated / dual-MI grid | val_fit · 39 configurations | 0.771331 vs 0.770556 · +0.000775 | BELOW INTEGRATION GATE · original MI1024 retained |
| CMY + RT-DETRv4 + Tiny/ViT | reported grouped80 experiment | 0.763811 vs 0.762515 · +0.001296 | REJECTED GLOBAL MEMBER · no hidden result |
| Tiny384 letterbox + ViT-S | reported grouped80 experiment | 0.764970 vs 0.762515 · +0.002455 | BELOW INTEGRATION GATE · no hidden result |
| LambdaRank scene context | fixed w=0.7 | 0.775815 vs 0.776482 · -0.000666 | REJECTED · unchanged ranker retained |
| PAST · completed evidence | FUTURE · unresolved work |
|---|---|
| grouped80 freeze · 12-model 1024 audit · MI-DETR1280 E13 · RT-DETRv4-X1280 · CMY ensemble · Tiny/DINOv3 rescorer · LambdaRank · owner-verified 0.7910702 · missing-annotation audit · TRAIN 3/3 pool · diagnostic views · synthetic V4 freeze · rare tuning · ignore pilots · rescorer v2 handoff · U812 handoff | teammate rescorer retraining · canonical and PRIMARY diagnostic evaluation · U812 complementarity test · integrate only validation-proven changes · exact 0.7910702 repository synchronization |
MI-DETR1280 Effective-Scale V2 E13
E13 / update 9836 is accepted at local 0.7258748594 and AIdea 0.7373246. E14 was lower and E15 was not run.
Rescorer retraining data v2
Teammate training result is pending. The package uses protected negative exclusions, not pseudo-positive GT.
MI-DETR U812
Local regression is not promoted. Test only as an additional complementary detector with raw float predictions.
Unresolved evidence · last documented work
1 · Preserve accepted evidenceMI-DETR1280 E13 remains the best accepted standalone; E14 is lower and E15 was not run
2 · Evaluate rescorer v2Report canonical and PRIMARY diagnostic evaluation; do not promote without a result
3 · Test U812 complementarityUse raw float predictions as an additional detector only; it does not replace original MI-DETR1024
4 · Integrate only proven changesProtect the team pipeline from unverified changes
5 · Synchronize 0.7910702 lineageBind exact artifact and pipeline only when repository evidence matches the owner-verified leaderboard result
| Variant / branch | Current state | Next sufficient action |
|---|---|---|
| MI-DETR1280 Effective-Scale AUG V2 E13 | COMPLETE · ACCEPTED · local 0.7258748594 · AIdea 0.7373246 | preserve as best accepted standalone |
| Co-DINO1280 Full-Aug V1 | FAILED · SUPERSEDED | retire; do not continue |
| Rescorer retraining v2 | HANDOFF READY · result pending | run canonical and PRIMARY diagnostic evaluation |
| MI-DETR U812 | CANDIDATE PENDING HIDDEN · local regression | empirical complementarity test only |
| 0.7910702 team result | OWNER-VERIFIED | repository lineage synchronization |
| Pseudo-positive experiment | NOT STARTED | do not present as approved or completed |
strong model completion → ensemble optimization → crop rescoring → final submission system
| Item | Current authority |
|---|---|
| Component | ConvNeXt-Tiny crop rescorer · PROVEN HIGH-VALUE COMPETITION COMPONENT |
| Observed evidence | 0.7701133 AIdea · team pipeline · historical 90,393 rows |
| Fully controlled fallback | Co-DINO1280 + MI-DETR1024 WBF · local 0.7468375071854423 · hidden 0.7587924 |
| Exact-reproduction boundary | local candidate 90,391 rows · tiny threshold-boundary difference · not a replacement |
| Artifact convention | small provenance in Git; large prediction/checkpoint artifacts in LFS/shared storage; datasets externally |