This dataset is released for research use. Access is reviewed and granted manually by the maintainers. Please state your name, affiliation, and intended use.

D23-mcq

Marked-instance multiple choice on VISION industrial imagery — 140 records (one balanced-picked marked instance per eligible record, drawn from an eligible pool of 384 marked instances), derived deterministically (no LLM/teacher) from the human gold boxes of AI4Manufacturing/D23-annotated. Exact-match gradable (single letter) → SFT and RLVR-ready.

The repository name is an internal task code. See Provenance below.

Smallness is deliberate. 140 records over 2 subsets is what survives the eligibility and kill battery below. The rung ships anyway for format parity: the corpus PoC found the MCQ shape own-task-critical and probe-protective, and the letter protocol here matches the D15/181 MCQ rungs so per-shape findings transfer.

Query diversity. query is drawn from a pool of 32 surface variants (paraphrases preserving task and answer format; the options clause and answer-format directive are held verbatim), selected by an independent salted per-item hash — the corpus query-template-diversity standard.

Task

One defect instance is marked on the full-frame image by a red rectangular outline (same drawing + seeded per-side [10%, 40%] padding convention as AI4Manufacturing/D23-region); the query lists the subset's class menu as lettered options and asks for the letter only. Position-fair seeded letters: the class→letter assignment is an independent salted-hash permutation per record, so no letter position correlates with the answer. Gold letter counts: Lens A 15, B 18, C 19, D 11, E 21; PCB_1 A 7, B 14, C 8, D 8, E 9, F 10.

No no defect option — every record's marked instance is a real gold, so the option would be never-correct and is excised (corpus-wide surgery; it exists only in D23-region's PCB_1 menu, where decoys make it correct).

Balanced instance pick. Each record contributes ONE marked instance, chosen by the gate battery's balanced per-record pick (least-picked class first, deterministic order) over its 384-instance eligible pool — this caps the answer-class priors at the shares below (shipping the full pool would breach the prior cap).

Records

140 records (train=71 · validation=69), split preserved verbatim from the parent.

subset	records (train/val)	eligible instance pool	options	max answer-class share	max gold-letter share
Lens	84 (41/43)	218 (100/118)	5	26.19%	25.00% (uniform 20.00%)
PCB_1	56 (30/26)	166 (84/82)	6	42.86%	25.00% (uniform 16.67%)

Every menu class is gold ≥1× within its subset. PCB_1 spur note: spur has exactly 1 gold pick (train-side only) — on the validation side spur is an eval-side distractor-only option.
Blind-guess ceiling = the max answer-class share above (letter-marginal guessing is capped by the letter fairness column).

field	type	meaning
`query`	str	names the product, mentions the red outline, lists lettered options; "answer with the letter only"
`image`	Image	full-frame photo (never cropped) with the padded red outline burned in
`annot`	str	gold letter (exact match)
`reasoning`	null	none — deterministic derivation
`cate` / `task`	str	`B` / `T-B2` (unified schema; rungs keep the parent token)
`metadata`	str (JSON)	subset, split, `image_sha256`, `source_record_id`, `instance_index` (deduplicated objects list), `gold_class`, `gold_letter`, `letter_map`, raw `bbox_xywh` + `padded_box_xywh` + `drawn_box_xywh`, menu, gates provenance

Eligibility & kills (the decision trail)

Pools were frozen by a pre-build gate battery under an adversarial convergence review; the numbers below are the battery's own measurements.

Legibility floor. An instance is eligible only if min(w,h) ≥ 16px at the 2.36MP-equivalent training resolution. Unlearnable fraction of each shipped subset's parent instances by reference input (long side / area cap):

subset	448	768	1024	1568	2.36MP
Lens	93.4%	87.8%	83.5%	68.1%	68.8%
PCB_1	100.0%	97.5%	85.6%	69.7%	58.1%

Geometry-decoder kills (rule: hard decoder iff probe >= majority + 25pts AND >= 75% absolute, computed on padded outline geometry — no retained subset's outline geometry decodes its answer):

PCB_2 killed — full-geometry GBM 98.1% vs 30.5% majority (repeated-panel layout).
Cable killed (region/MCQ) — size-only GBM 82.4% vs 50.4% majority: box size alone decodes the class.
Casting killed — full-geometry GBM 83.8% vs 56.6% majority; the kill is position-driven (position channel adds +10.0pts over size-only 73.8%).
Cylinder killed (round-3) — cross-split geometry twins: 21.7% of val instances have a train instance within 10px (raw geometry) and 51.3% within 30px; train→val padded 1-NN 80.3% vs 33.6% val-majority — repeated rig positions leak labels across the split.

Retained-subset certifications (same probes, below the bar):

Hemisphere (BCD) — cross-split padded 1-NN 71.9% (+16.7pts over val-majority) with twin fractions 0.0%/8.1% — the signal is class-geometry, not rig repetition; in-pool padded LOO 1-NN on the shipped pool = 0.773 vs kill bar 0.780 (passes by 0.7pts; both digits disclosed — this is the highest retained value in the corpus).
Lens (post-Fiber-trim) — cross-split padded 1-NN 50.0% (-8.0pts), twins 0.9%/3.6%.
PCB_1 — cross-split padded 1-NN 37.8% (-26.8pts), twins 3.7%/7.3%.

Opaque-code discriminability (size-normalized audit). Hemisphere's anonymized Defect-A..D codes were additionally required to show appearance-borne signal on a fixed 224px canvas (removes absolute size by construction; judge-side instrument, upscaling allowed): Defect-B 3.2× chance, Defect-C 2.0×, Defect-D 1.73× pass the ≥1.5× bar; Defect-A 0.8× fails and is excluded as a gold everywhere (menu excised too — never-correct options are removed corpus-wide). The verdicts are judge-robust (judge-sensitivity rerun, three control-passing judges: claude-sonnet-4-5, claude-sonnet-4-6, gpt-5.6): Defect-A stays under the ≥1.5× bar with every judge (accuracy 0.20/0.37/0.30 vs the 0.375 bar), while Defect-B/C/D and every PCB_2 class pass under all three. Instrument-ceiling caveat: the positive control (Cylinder, semantically named classes) passes the control gate under every judge (macro 0.65/0.74/0.75), but its Chip class scores 0.15 under the original judge and rises to ≈0.45 under the stronger judges — per-class collapses are partly judge-limited, not a hard pixel ceiling; treat per-class passes as a lower bound on discriminability. Residual cue disclosed by the protocol: aspect ratio survives the canvas normalization.

Additionally for MCQ: Hemisphere is excluded from this rung (choosing among opaque Defect-* codes from a menu is closer to memorization than perception; audit scope), independent of the region-rung retention above.

No D23-counting rung exists: counting was dropped at plan review (degenerate ~all-one count priors on this parent + the counting shape lacks PoC validation; the shape is kept corpus-wide via other artifacts).

Roles

Roles: this is an answer-only tier — there is no reasoning content (reasoning is null on every record); annot is both the machine-parseable gold AND the direct-answer SFT target ('SFT-ready' here means direct imitation of annot in the query-specified format); it is also the exact-match/IoU reward key for RLVR.

Provenance

Derived read-only from AI4Manufacturing/D23-annotated (revision pinned: fd728cc34406cd87512bd22fd8c31df78dace149), itself derived from raw VISION (Bai et al., arXiv:2306.07890; upstream CC BY-NC 4.0 — respect upstream terms; this card is license: other). Answers are pure functions of the human gold annotations — no LLM/teacher anywhere in this rung. Generator: annotate/D23/rungs/build_rungs.py in AI4Manufacturing/forge_model (builder sha256 8652395b…), built against the pre-build gate battery (gates v5 report, script sha256 0b9d0e43…, build source of truth = its section 16). Every independent choice is salted-hash seeded; a rerun reproduces the artifact exactly (tuple-hash verified).

Parent golds were deduplicated first: 5 records carried 8 exact-duplicate (class,bbox) phantom instances, collapsed before any pool math. metadata.instance_index refers to this deduplicated objects list, not the raw parent array. The parent's official train/validation split is preserved verbatim on every item (uniform-split policy; carve train/eval downstream).

Training-mixture notes

One-lineage rule: the parent excludes the 641 VISION images byte-shared with the DefectSpectrum/D15 family (materially different label policies); the 8 subsets here have zero image overlap with the D15 family. Details + machine-readable keys: base AI4Manufacturing/D23 card §8.
Same-evidence rungs: 137 of 140 D23-mcq marked instances also ship as D23-region records with identical padded outlines — region and mcq are format variants over the same evidence. Treat them jointly in any mixture/eval carve; never split the two rungs across train/eval.
Image-wise carving: one photo can appear in the parent and in up to three rungs; carve on metadata.image_sha256 across the whole D23 family simultaneously.
All-defective world prior: every parent record is defective; rung items never assert a defect-free image (no 'no defect' option in this rung).

Overlap / de-duplication (§8)

Base photos are the SAME images as AI4Manufacturing/D23-annotated and base AI4Manufacturing/D23 — this rung inherits their overlap situation: the shipped 8-subset lineage shares no imagery with the D15 family (sha-verified sidecar on the base card), and D23-validation imagery appears in the other D23 rungs and the parent. Do not evaluate on any D23-family repo's validation split if you train on this set, and reconstruct exact overlaps via metadata.image_sha256.

Cross-family evaluation lock — metadata.eval_lock (stamped 2026-09-20; manifest revision fe6e286912b0, generated 2026-09-08). Every record of this repository, locked or not, carries metadata.eval_lock, computed by forge_model/common/overlap.py::Overlap.stamp_for against common/overlap_manifest.json at that revision — so within this repository the absence of the key cannot occur. Shape: {"locked": bool, "against": [{"repo": …, "split": …}, …], "own_split": …, "manifest_revision": …, "manifest_generated": …}. locked is true when the image is evaluation material anywhere in the corpus; against names every repository and split in which it is (sorted; [] when not locked; it includes the record's own family where that is so); own_split marks a record locked by its own split. The per-record field is the authority — the count here is quoted once, at this revision, and a later manifest may change it: 69 of 140 records (69 distinct images) are locked — by column: 0 by the cross-family manifest, 69 by their own split, 0 both ways and counted once; counterparts (records per counterpart; a record can appear under several): none — every lock here is by the record's own split; 69 locked by their own split: validation. In words: 69 of the 140 records in this repository are evaluation material by their own metadata.split (validation: 69) and sit inside the HF split named train / validation — under the uniform-split convention the HF split name is a container name, and metadata.split together with metadata.eval_lock carries the truth; a train pool must exclude them. A stamp whose manifest_revision differs from the current manifest is stale, not wrong — recompute it (Overlap.stamp_is_current); a record with no stamp has not been checked against the corpus as it now is. Overlap.partition / assert_train_pool_clean read the field: a train pool built from this repository must exclude every locked record.

Training notes

annot is a single letter → loss on completions; exact-match reward for RLVR.
137/140 of these marked instances also ship as AI4Manufacturing/D23-region records with identical outlines — same-evidence format variants; never split the two rungs across train/eval.
Companions: AI4Manufacturing/D23-annotated, AI4Manufacturing/D23-region, AI4Manufacturing/D23-grounding.

Geometry (`metadata.geometry`)

Every record carries a geometry block inside the existing metadata JSON string, so that its gold can be re-derived at any render size. No schema column changed; existing loaders are unaffected.

Coordinates are native pixels of the image in that record (coords_frame: "record_image"). scale is 1.0 throughout — this repo publishes at source resolution, nothing was downscaled at publish time.

"geometry": {
  "image_wh":  [W, H],        // dims of the image in THIS record
  "source_wh": [W, H],        // dims of the original source image
  "scale": 1.0,               // image_wh / source_wh; < 1.0 would disclose a publish-time downscale
  "n_instances": 2,
  "instances": [
    { "instance_id": 1, "bbox_xywh": [x, y, w, h], "min_side_px": 65, "class": null }
  ],
  "n_dropped_subminimum": 0,  // components removed by the filters below
  "union_box_fallback": false,// true => boxes are per-class unions, NOT real instances
  "conventions": { ... }      // see table
}

instances is present even when empty. [] means the record genuinely has no defects; an absent block would mean geometry could not be recovered. Those are different states and are never conflated.

Conventions used to derive it

There is no universal definition of "one defect instance" — it depends on the mask the source shipped. This repo's is stated, not implied:

field	value
`algorithm`	`source_annotation`
`binarisation`	`n/a`
`connectivity`	`4`
`merge`	`none`
`min_area_px`	`0`
`max_instances`	`None`
`artifact`	`fine`
`fill_floor`	`None`
`legibility_floor_px`	`None`
`min_side_floor_px`	`None`
`spec_sha`	`a98e8ed866091f30`

Provenance and verification


records	140
carrying a geometry block	140 / 140
instances per record	`1`: 16, `2`: 12, `3`: 18, `4`: 5, `5+`: 89
total instances	869
image dimensions	2448×2048 (24), 512×512 (24), 2904×1921 (11)
`scale` values present	[1.0]

Derived from the AI4Manufacturing/D23 masks and verified against this repo's own published answers before it was written — a recomputation that disagreed with the shipped gold would have aborted the update rather than overwritten it.

⚠ The 16px floor applies at the RENDER, not at native

min_side_px is in native pixels. The model does not see native: Qwen2-VL caps by megapixels AND snaps each dimension to a multiple of 28. So min_side_px >= 16 is the floor tested in the wrong frame. Measured on this repo:


native → rendered (qwen2_vl @ 2.36MP)	512×512 → 504×504, 2240×2016 → 1596×1456, 2282×2248 → 1540×1512
shipped boxes	869
legible at that render (>=16px there)	371 (42.7%)

⚠ An earlier version of this section reported the inverse — boxes clearing 16px at native and failing at the render — and that number was misleading. It is frame-relative: publishing at a larger native size lets more boxes clear 16 in the published frame, so more can "fail", which penalises exactly the choice that helps. Measured on 179: publishing native (3024) means a box needs >=32px native to be legible at the render and 86.7% qualify; the previous 1024 publish needed >=47px native and only 69.5% qualified. The native republish improved rendered legibility by 17 points while the old metric scored it as 12.5% "broken". The figure above is the comparable one.

Nothing in the data is frame-dependent — geometry is native and complete. Use forge_model/D23/adapt.py, which applies the floor at whatever size the consumer renders.

Using it

Coordinates only stay correct if they are rescaled with the image. A patch-based VLM does not render at native size: Qwen2-VL's processor snaps both dimensions to a multiple of 28, so this repo's 512×512 is rendered 504×504 and native-pixel boxes are then wrong by a few pixels. forge_model/D23/adapt.py regenerates coordinates for a target render size, re-derives counts, and drops records whose gold no longer holds there.

Montage render tax — not applicable (2026-09-19). Each record's image is the single full source frame (geometry.image_wh == source_wh, coords_frame: record_image, scale 1.0) with one red rectangular outline; there is no 2×2 composite here, so no montage tax applies and the legibility figures above are single-frame figures. (Stated so the three mask-MCQ cards read alike: D15-mcq and 181-mcq are composites and carry a measured tax.)

Downloads last month: 85

Total file size:

558 MB

Paper for AI4Manufacturing/D23-mcq

VISION Datasets: A Benchmark for Vision-based InduStrial InspectiON

Paper • 2306.07890 • Published Jun 13, 2023 • 1

You need to agree to share your contact information to access this dataset