Hand-Quality Labelling · v0.31
(Last edited · 04 Aug 2025)
Why this matters Every accurate label you create nudges our generative-image models closer to anatomically believable hands. Thank you for being part of that.
1 Purpose
Modern diffusion and transformer models still struggle with hands: wrong finger counts, melted geometry, odd proportions. We’re building a high-quality dataset—cropped hand images with human judgements of “passable vs. defective.”
Your labels will:
-
Train automatic defect detectors (pre-filtering bad hands).
-
Feed research into anatomy-aware generation.
Short version – judge the hand, not the whole artwork; judge severity, not resolution.
2 Rating Design
2.1 Questions (unchanged)
| Q# | Prompt | Allowed answers | Notes |
|---|---|---|---|
| 1 | How would you rate the quality of this hand? | Good, Bad, Not Sure, No Hands | Not Sure ≤ 5 % overall. |
| 2 | If Bad, why? | Severely Deformed · Missing⁄Extra Digits · Subtle Distortion · Other | Multi-select disabled—pick the single worst defect you see. |
2.2 Ground Rules
-
Severity first. A tiny, far-away hand that is clearly fused is still Severely Deformed.
-
One bad hand → bad image. If any visible hand fails, the whole crop is Bad.
-
Ask, don’t guess. When torn, choose Not Sure and ping @Yada in hands-labelling.
3 Rating Examples – Quick Lookup
| Category | Concise definition | Image placeholders |
|---|---|---|
| Good | Five digits, plausible shape, no obvious artefacts. | GOOD-01 … |
| Bad — Severely Deformed | Fused, melted, broken, impossible joints. | SEVERE-01 … |
| Bad — Missing⁄Extra Digits | Looks human, but digit count ≠ 5 or ghost digits. | DIGIT-01 … |
| Bad — Subtle Distortion | Five digits yet odd proportions / “noodle” fingers. | SUBTLE-01 … |
| Bad — Other | Rare defects that fit none of the above. | OTHER-01 |
| Not Sure | Ambiguous after a second look. | NS-01 |
| No Hands | No hand visible at all. | NOHAND-01 … |
4 Rating Workflow
[Flow-chart to be inserted] — text version below.
-
Hands present? • No → No Hands • Yes → Step 2
-
Severe deformity visibly present? • Yes → Bad – Severely Deformed
-
Digit count off? • Yes → Bad – Missing⁄Extra Digits
-
Five digits yet anatomy “off”? • Yes → Bad – Subtle Distortion
-
Otherwise → Good
-
Still unsure at any point → Not Sure and leave a comment.
(Tip: zoom; don’t upscale. If blur alone hides the anatomy, choose Not Sure.)
5 Ambiguity Handling – Mini FAQ
| Scenario | Guidance |
|---|---|
| Partial hand occluded | If visible part shows a clear defect → label Bad; if hidden → Not Sure. |
| Stylised / cartoon hands | Ask whether stylisation keeps five distinct, proportionate digits. If yes, possibly Good. |
| Multiple hands in crop | Judge the worst hand. |
| Motion blur / low-res | Blur hides anatomy → Not Sure. Defect still obvious → relevant Bad category. |
| Edge of frame | Cropped finger counts as “missing” only if clearly absent, not merely out-of-frame. |
(Add visual edge-case call-outs when available.)
6 Examples of Past Mis-Labelling
“Crisp ≠ correct.” In the pilot, many crisp, painterly hands were marked Good despite elongated or fused fingers. Remember: looks-human ≠ anatomically-sound.
| Common error | Correct label | Why |
|---|---|---|
| Sharp, five-digit hand but fingers twice normal length | Bad – Subtle Distortion | Proportions matter. |
| Human-drawn but fused thumb-index | Bad – Severely Deformed | Fusion overrides artistry. |
| Cartoon mitt with three fingers | Not Sure (or Bad – Other if policy set) | Lacks standard anatomy; flag. |
(Insert illustrative crops.)
7 Version History
| Ver | Date | Changes |
|---|---|---|
| 0.1 | 13 Jul 2025 | Initial clean-up. |
| 0.2 | 27 Jul 2025 | Added quick table; trimmed prose. |
| 0.3 | 01 Aug 2025 | Pilot feedback; ambiguity notes. |
| 0.4-scratch | 04 Aug 2025 | Polished structure; workflow, mini FAQ, mis-label table; placeholders added. |
Next actions
-
Fill blanks: insert flow-chart and final image examples.
-
Run a 100-image calibration batch—track where Not Sure is used.
-
Update FAQ / examples, then bump to v0.4 final.
Questions, edge cases, or uncertainty? → Post in hands-labelling or tag @Yada.