Issues

• Tell a head whose positives sit under the bulk's tail before shrinking its count (Binary Photo at 1%)

  claudeexperiment
  samggreenberg/VTSearch#5215 · Filter by author samggreenberg opened on Oct 25, 2026

• Re-price 0.71 beta on Region Photo with more seeds and the full 150 clicks once the V100s free

  claudeexperiment
  samggreenberg/VTSearch#5214 · Filter by author samggreenberg opened on Oct 25, 2026

• SAM3

  enhancementexperiment
  samggreenberg/VTSearch#5213 · Filter by author MatthewELucio opened on Oct 23, 2026

• launchChromium's fallback misses Playwright's chrome-linux64 layout

  claudesolved
  samggreenberg/VTSearch#5208 · Filter by author samggreenberg opened on Oct 23, 2026
  1

• Slides: redraw the remaining blue threshold lines as dotted black

  claudesolved
  samggreenberg/VTSearch#5206 · Filter by author samggreenberg opened on Oct 23, 2026
  1

• New Detector on a Face dataset can't take its example from the loaded faces

  claudesolved
  samggreenberg/VTSearch#5205 · Filter by author samggreenberg opened on Oct 23, 2026
  1

• Slides

  solved
  samggreenberg/VTSearch#5203 · Filter by author samggreenberg opened on Oct 23, 2026
  1

• Rebuild coco_better with the #4954 label corrections, and settle its two open control rechecks

  claudeexperimentsolved
  samggreenberg/VTSearch#5191 · Filter by author samggreenberg opened on Oct 23, 2026
  1

• The head's count of what stands out counts the bulk's heavy tail: Region reads 2.6x its positives, and Binary's best multiplier moves with prevalence

  claudeexperimentsolved
  samggreenberg/VTSearch#5084 · Filter by author samggreenberg opened on Oct 18, 2026
  1

• Price the weak-separation check on COCO at today's head: Binary and the Region path

  claudeexperiment
  samggreenberg/VTSearch#4797 · Filter by author samggreenberg opened on Oct 12, 2026

• New SotA option for comparing Region voting approaches

  experiment
  samggreenberg/VTSearch#4711 · Filter by author MatthewELucio opened on Oct 9, 2026
  

• Tiled Stage 1: the 2,000-page shortlist loses positives as the collection grows (9.7-13% unverified at 200k pages)

  claudeexperiment
  samggreenberg/VTSearch#4493 · Filter by author samggreenberg opened on Oct 4, 2026
  6

• Lift the pandas<3 pin once the GPU stack can take RAPIDS ≥ 26.8

  claude
  samggreenberg/VTSearch#4390 · Filter by author samggreenberg opened on Sep 30, 2026
  4

• Born-digital benchmark: a curated screenshot dataset with a screen-genre track and an exact-icon track

  claudeexperiment
  samggreenberg/VTSearch#4356 · Filter by author samggreenberg opened on Sep 30, 2026
  7

• Size vs size on Region Photo: does region voting close the small-object gap?

  claudeexperiment
  samggreenberg/VTSearch#4217 · Filter by author samggreenberg opened on Sep 28, 2026
  1

• Structural templates from Good votes pick up printed text: keep the crop's template apart, or mask text

  claudeexperiment
  samggreenberg/VTSearch#4170 · Filter by author samggreenberg opened on Sep 24, 2026
  8

• Pin MAX_STRUCTURAL_DETECT_PIXELS on a real retrieval corpus, not a synthetic warp sweep

  claudeexperiment
  samggreenberg/VTSearch#3908 · Filter by author samggreenberg opened on Sep 16, 2026
  1

• How much does one wrong label early in a run change its outcome? (vg_scale path-sensitivity)

  claudeexperiment
  samggreenberg/VTSearch#3688 · Filter by author samggreenberg opened on Sep 6, 2026
  2

• Find out who uses the autorun extractor/localizer surface before touching it

  claude
  samggreenberg/VTSearch#3452 · Filter by author samggreenberg opened on Sep 1, 2026

• ROI Align

  experiment
  samggreenberg/VTSearch#2890 · Filter by author samggreenberg opened on Aug 7, 2026
  1