We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
docs: detection 2026-04 페이지 이미지 placeholder 5종 실제 차트로 교체 - progression_chart.png: stage별 mAP50/mAP50-95 line - stage_grouped_bars.png: 그룹별 막대 비교 - epoch_trajectory.png: yolo11m@1280 vs @1536 val 추이 + test 표시 - 보조 차트 추가: - axis_effects.png: 각 변경 축의 Δ mAP50-95 - pr_scatter.png: Recall × mAP50-95 산점도 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
docs: detection progression 2026-04 페이지에 후속 작업 섹션 + yield 파이프라인 링크 추가 - Fair chain 4-way 결과 (yolov8m 0.286 / yolo11m 0.291 / yolo11l 0.284 / yolo26m 0.283)로 plateau 확정 명시 - 후속 작업 섹션 추가: detection axis 종료 → segmentation axis 전환 - 수확량 예측 파이프라인 (2026-05) 페이지로 링크 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Updates progression doc with Stage A and Stage B2 results - Stage A (yolo11m@1280): mAP50-95 0.286 → 0.257 (-0.029, missed) - Stage B2 (yolo11m@1536): mAP50-95 0.286 → 0.251 (-0.035, missed) - Adds dedicated S2 → B2 interpretation section with 4 hypothesis table - Adds combined A+B2 retrospective: "model/resolution axis is blocked, pivot to data axis" - Updates next-candidates table: B2 strikethrough, oversample ratio sweep / additional drone labeling raised to top priority - Updates retrospective: adds 4th item on simultaneous "scale-up" axes overfitting risk
Adds drone detection progression document and links from related pages - Drone-Detection-Progression-2026-04.md: stage-by-stage progression with mAP results, decision rationale, and retrospective on what worked vs missed - Home.md: adds the new progression doc to the recommended reading order for the grape project - 포도밭-병해충-탐지-및-수확량-예측.md: adds new "학습 변천사와 의사결정 회고" section linking to the progression doc