Skip to content

v0.2.0 — Phase 2: Template Generation via Gemini 2.5 Pro

Choose a tag to compare

View all tags
@MuratAlkan06 MuratAlkan06 released this 09 May 02:30
· 6 commits to main since this release
v0.2.0
5e6e13b
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

What is in this release

Phase 2 of EasyLab is complete. The application can now take a user-annotated reference image and generate a semantic template by calling Gemini 2.5 Pro, which enriches each annotated field with a semantic_description stored in template_fields.

Changes

Worker (FastAPI):

  • Stage A pipeline: EXIF-transpose preprocessing, Gemini 2.5 Pro call with response_json_schema via Pydantic, retry with tenacity, degraded fallback when Gemini fails twice
  • Spend controls enforced before any Gemini call (workspace_quota check + global token budget)

Web (Next.js):

  • API routes: project CRUD, image upload (signed URLs), image confirm, reference image selection, field management, job enqueue, job status polling
  • Pages: Dashboard, Upload, Annotate (react-konva canvas), and Process (polling + status display)

Phase boundary

Closes the Phase 2 milestone. Phase 3 (full detection + extraction, review table population) is next.

Assets 2
Loading