WI-038 · Research & Discovery · research-corpus-pipeline

Build tagging corpus from gathered photo sets

queued P2 low risk   Owner: Paul   Reviewer: Paul   0% · 0/2 tasks complete

Gates: R1 R2 Requirements: REQ-RESEARCH-001 REQ-RESEARCH-002 Updated: 2026-06-22

At a glance

CodeWI-038
PhaseResearch & Discovery
Order73 of 93
Story points5
Primary surfaceCorpus: organize + tag the gathered garage/estate photo sets for vision reference/training
Retires
Depends on
Blocks
Evidence
pending

Goals

Build tagging corpus from gathered photo sets — advances R1, R2 toward its gate(s).

Implementation — what to build

Intake

  • Ingest the gathered photo sets (HITL-RESEARCH-001/002) from Storage.

Organize + tag

  • Tag images by category / maker-mark / condition into a vision reference/training corpus (kept SEPARATE from canonical inventory).

Tooling

  • Simple tagging UI (Paul's task) to label the gathered images.

Use

  • Seeds vision benchmarking (local-vs-hosted), watchlist clues, and taxonomy validation.

Acceptance

  • Gathered sets are organized + tagged + queryable as a corpus; never canonical inventory.

Checklist

  • R1 Garage-sale field corpus: consistent sample photo sets + item samples collected for reference and future vision training.
  • R2 Estate-sale field corpus: consistent sample photo sets + item samples collected for reference and future vision training.

Gates & testing

  • R1 Garage-sale field corpus: consistent sample photo sets + item samples collected for reference and future vision training.
    HITL QA
  • R2 Estate-sale field corpus: consistent sample photo sets + item samples collected for reference and future vision training.
    HITL QA

Requirements

  • REQ-RESEARCH-001 proposed Collect a garage-sale field corpus: consistent sample photo sets + item samples for reference and future vision training.
    Done when: Sample sets uploaded with consistent capture (scene + item shots), reviewed and accepted by Paul; usable as vision reference/training seed.
    Spec: Human direction 2026-06-19 · docs/CC-Analysis/trovesnap-vision-scanning-spec.md (28.10 internal corpus)
  • REQ-RESEARCH-002 proposed Collect an estate-sale field corpus: consistent sample photo sets + item samples for reference and future vision training.
    Done when: Sample sets uploaded with consistent capture, reviewed and accepted by Paul; usable as vision reference/training seed.

Verify (done when)

the photo sets gathered by Justin (HITL-RESEARCH-001/002) are organized and tagged by category into a queryable corpus for vision reference/training; never canonical inventory.

Cost triggers

No paid API / infra spend triggered by this item.

Status & editing

queued   Sprint: Research & Discovery

Edit status / sprint on the ★ Live Board → — changes are logged live with who / when / why.

Human-in-the-loop

HITL-RESEARCH-001: Gather garage-sale photo sets waiting
Justin Hilleary · waiting on Justin Hilleary
HITL-RESEARCH-002: Gather estate-sale photo sets waiting
Justin Hilleary · waiting on Justin Hilleary

Findings

None recorded yet.

Concerns

None recorded yet.

Risks

None recorded yet.

Tech debt

None recorded yet.

Tools used

No tool calls recorded.

Files & artifacts

No files / artifacts recorded.

Update log

No log entries yet.