Commit History

  1. submission: add hackathon track/sponsor/achievement tags to README frontmatter

    Apiarist Dev
    9e0d994
  2. docs: add submission section (track, demo video, social post, badges); soften cost claims

    Apiarist Dev
    50afe3b
  3. fix: PDF notes wrap inside table; hide trend chart until 2+ inspections (no empty box)

    Apiarist Dev
    b378fd6
  4. chore: remove internal planning docs (drafts, scripts notes) from public repo

    Apiarist Dev
    3bf4d3f
  5. fix: comparison report sentence built from specialist counts (no more mite contradiction)

    Apiarist Dev
    be50c21
  6. ui: fill_width=True so layout uses full width (capped at 1600px via css)

    Apiarist Dev
    e4f30a9
  7. ui: queen tile shows count (1/0) not YES; widen layout to 1600px

    Apiarist Dev
    c11bda6
  8. chore: trigger rebuild to regenerate frontend assets (cache-bust)

    Apiarist Dev
    8987cc2
  9. off-brand: custom tab bar, themed inputs/labels/tables, custom scrollbar, strip Gradio tells

    Apiarist Dev
    dc2bfb3
  10. feat: inspection dashboard (verdict + stat tiles), colony trend charts, polished A/B kill-shot cards

    Apiarist Dev
    be206e5
  11. polish: hero header + detection legend + footer (Off-Brand); document Modal training pipeline + published artifacts

    Apiarist Dev
    34483f7
  12. samples: add verified mite-detection and drone-detection photos (proves all 4 classes)

    Apiarist Dev
    0fecff4
  13. remove cyan queen candidate boxes - only confident green QUEEN box now

    Apiarist Dev
    8ec418b
  14. fix: replace both collage samples with clean single-scene hive photos (collage detector verified)

    Apiarist Dev
    1327e19

  1. fix: replace collage sample photo with single-scene queen close-up

    Apiarist Dev
    b5ea41e
  2. samples: 2 queen+drones + 3 queen-only photos, all verified working

    Apiarist Dev
    f43b366
  3. feat: test-time augmentation (4 views) + size filter (top 30% bbox area) for queen detection

    Apiarist Dev
    85cec4b
  4. feat: ranked top-3 queen candidates with probability shown - honest AI, user verifies

    Apiarist Dev
    7119114
  5. tune: raise queen threshold 0.85 -> 0.92 to reduce false positives on moderate-distance shots

    Apiarist Dev
    9af6826
  6. fix: dynamic inference size (640/1280/1920 by image dimensions) - small thumbnails were being upscaled too aggressively

    Apiarist Dev
    48dd007
  7. tune: imgsz=1920 + lower thresholds to catch more bees on high-res user photos

    Apiarist Dev
    ffbb2d5
  8. feat: 5 sample photos with verified queen detection + disclaimer about operating conditions

    Apiarist Dev
    e6c0dcb
  9. feat: ship trained queen classifier (F1=0.962) - 100% confidence on real queen, near-zero on negatives

    Apiarist Dev
    c535f2c
  10. feat: binary classifier cascade - prefers dedicated queen classifier, falls back to VLM grid

    Apiarist Dev
    b27c5e3
  11. fix: cascade no longer rubber-stamps YOLO queen labels - decides fresh from size only, ultra-strict parse

    Apiarist Dev
    2e7ab44
  12. fix: cascade prompt emphasises color + size, more candidates, lower bee threshold

    Apiarist Dev
    9a02a0e
  13. fix: cascade is too permissive - require exactly one queen, stricter prompt

    Apiarist Dev
    31fa0f9
  14. feat: VLM cascade for queen verification (YOLO finds candidates, Qwen-VL confirms via grid comparison)

    Apiarist Dev
    ae234b4
  15. feat: lower queen threshold to 0.05, show top-confidence per class in narrative

    Apiarist Dev
    0111bb2
  16. feat: ship combined-dataset YOLO weights (Matt Nudi + Hendricks; queen mAP 0.99 + better generalization)

    Apiarist Dev
    df673a9
  17. fix: distinguish empty YOLO detections from missing weights in narrative

    Apiarist Dev
    f4f2593
  18. fix: distinguish empty detections from missing weights; lower per-class thresholds for retrained model

    Apiarist Dev
    1340369
  19. chore: nudge build hook

    Apiarist Dev
    49524d4

  1. feat: ship retrained YOLO weights (queen mAP 0.992) + smarter class canonicalization

    Apiarist Dev
    e9abb16
  2. polish: remove emojis and em dashes from sources and docs

    Apiarist Dev
    238bdf6
  3. polish: seed sample hives, polished CSS theme, rich About + README, dataset live on Hub

    Apiarist Dev
    686d2ed
  4. docs: demo video script, blog post draft, LinkedIn post draft (Field Notes badge)

    Apiarist Dev
    87f44ca
  5. feat: class aliasing + varroa support, ready for hendricks_ricky weights

    Apiarist Dev
    d92374a

  1. fix: per-class confidence thresholds, queen-only label, mites clearly marked as VLM estimate

    Apiarist Dev
    e5b1be5
  2. fix: custom annotation — thin per-class boxes, only label queen+drone, no bee text

    Apiarist Dev
    582e81b
  3. fix: force YOLO onto CPU (ZeroGPU rejects CUDA ops outside @gpu )

    Apiarist Dev
    aa747ca
  4. refactor: YOLO runs on main container (not GPU worker), Qwen-only in @gpu

    Apiarist Dev
    03ce53c
  5. fix: run YOLO on main container, not GPU worker (file access + log visibility)

    Apiarist Dev
    19fed7f
  6. debug: log detector path/size at first call + retry instead of caching failure

    Apiarist Dev
    3f394fc
  7. feat: ship custom-trained YOLOv8s honey-bee/queen/drone detector (22MB)

    Apiarist Dev
    443d760
  8. feat: PDF weekly report generator + download button in Hives tab

    Apiarist Dev
    c87dd80
  9. feat: real A/B comparison logic (Apiarist vs Qwen-alone)

    Apiarist Dev
    4f7f771
  10. feat: YOLO + Qwen pipeline with graceful fallback when weights absent

    Apiarist Dev
    1032894
  11. feat: A/B Compare tab shell, ready for YOLO weights

    Apiarist Dev
    0381c4f
  12. fix: refresh Inspect tab hive dropdown when hives are added/refreshed/analyzed

    Apiarist Dev
    9ea6c2e