danial2026/demo-vision
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
No description
7 commits 1 branch 0 tags 930 KiB
Find a file
Exact
danial b6248f7234 scripts: add start-all.ps1 for the vision stack, docs pointer
2026-10-11 14:16:40 +03:30
app scripts: add start-all.ps1 for the vision stack, docs pointer 2026-10-11 14:16:40 +03:30
docs app v1 + e2e verdicts, idleTimeout fix, gallery 2026-10-11 01:09:06 +03:30
scripts scripts: add start-all.ps1 for the vision stack, docs pointer 2026-10-11 14:16:40 +03:30
test_files research vision shortlists + test files + repo skeleton 2026-10-10 22:48:29 +03:30
.gitignore research vision shortlists + test files + repo skeleton 2026-10-10 22:48:29 +03:30
README.md ui: drop tag emoji for TAGS prefix, refresh screenshots/docs, round 5 tests 2026-10-11 10:22:53 +03:30

README.md

demo-vision

Local vision-model bench for outfit description + visual Q&A, English and Farsi. Same B&W component look as demo-chat (separate repo); flows are image-centric: upload → describe → ask follow-ups → compare models.

Answered visual question with perf badge Sectioned system box, samples, Describe button Tag chips + collapsed tag prompts Compare chart + runs table Logs table

Layout (planned)

test_files/   E2E photos (user-supplied clothing pics + generated outfit-*.png)
docs/         research + setup notes
scripts/      downloads, eval sweeps
app/          bun + sqlite UI (describe, ask, compare)

Status

  • Research (docs/RESEARCH.md — vision shortlists for this PC)
  • Test files (test_files/: 5 user clothing photos + 3 generated outfits)
  • Model downloads (Ollama vision pulls)
  • E2E per model (EN describe + outfit Q + Farsi)
  • App (describe/ask/compare flows)
  • Screenshots, docs, remote push