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

Roles

Roles: canon repo — annot is the source label, kept machine-parseable as the gold for verification and reward parsing; there is no reasoning column and this repo is not itself a training view. Derived repos (-annotated, -grounding, -region, -mcq) each state their own regime on their own card. Geometry for every record lives in metadata.geometry (below).

D11

Factory/logistics scene object detection (COCO bbox; 5 classes). Category B, task T-B5b, in the unified Smart-Manufacturing SFT schema.

The repository name is an internal task code. See Provenance below for the underlying dataset.

Records

5,097 records (train=2820 · validation=2277).

Unified SFT schema

field	type	meaning
`query`	str	the question / instruction (model input)
`image`	Image	the input image (bytes embedded); for multi-image rows, a preview of the first view
`images`	list[Image]	(multi-image rows) all input views / modalities for the row, bytes embedded
`annot`	str	the answer — for this dataset: one line per detected object, `class,[x, y, width, height]` — exactly what the query asks for. Coordinates are NATIVE-pixel COCO xywh: top-left origin, [x, y, width, height] at the original image resolution — NOT xyxy, NOT normalized. Most grounding-capable VLMs use a different convention (norm-1000 xyxy, y-first PaliGemma/Gemini orders, Qwen2.5-VL resized-absolute); convert per student model before grounding-training — see `common/box_convert.py` in `AI4Manufacturing/forge_model`
`reasoning`	null	no native CoT in these datasets
`cate`	"B"	SFT category
`task`	"T-xx"	unified task id
`metadata`	str (JSON)	split, provenance, `image_path`, `image_sha256` (dedup key)
`mask`	Image \| null	(T-B1/T-B2 only) the pixel ground-truth mask, bytes embedded
`masks`	list[Image]	(multi-image T-B1 / D21) per-view masks aligned with `images` (None where a view has no defect), or multi-region masks

Query text — pooled paraphrases (v2)

Every record's query is drawn from common/vision_query_pools.json[D11/orig], a pool of 37 gate-verified paraphrases of the shipped wording, assigned by a stable hash of the source image path and recorded as metadata.query_template (37 templates in use, top share 3.1%).

Template 0 is v1's wording byte for byte (149 records keep it); the pass asserted that on every record before rewriting anything.

Template ↔ gold independence on this build: 5,097 records, 37 templates, worst template p = 0.000844, alpha 2.7e-04, 0 flagged → PASS.

Frame-size floor (common/lazy_floors.py, the standing (width, height)-only row): MAE 20.762 vs 20.406 blind median, 5 distinct frame sizes — no signal (5-fold cross-validation: the repository ships one split).

Answers, images, masks, split and every other field are byte-identical to v1: this revision was issued from the published parquet itself (tools/requery_published.py), not rebuilt from source, and the pixel-identity guard ran on the embedded images (§8 below).

Provenance

Underlying dataset: LOCO (tum-fml). Upstream license: CC0 1.0 (this card is license: other; respect the upstream terms). Converted read-only from the raw source into the unified schema; conversion code under D11/ (with publish/push_to_hf.py) in AI4Manufacturing/forge_model.

Overlap / de-duplication (§8)

Shares its image pool with the derived VQA set AI4Manufacturing/D11-QA-annotated (3,609 of these 5,097 images carry QA records; as of D11-QA v2, 2026-07-07, both datasets use the official LOCO v1 split, so train/validation sides agree — training on one set's train split never touches the other's validation images). Official LOCO v1 split (subsets 1&4=validation, 2/3/5=train); each record's source subset is in metadata.subset.

Two identities, and they answer different questions. metadata.image_sha256 hashes the file bytes: it finds byte-identical copies and is blind to a re-encode. metadata.pixel_sha256 hashes the decoded image (mode | size | pixels): it finds the same photograph saved twice. Only the second one settles whether an image is duplicated.

Measured at build time, not asserted afterwards — a violation aborts the build and names the offending records:


images checked	5,097
distinct by decoded pixels	5,097
images carrying more than one record	0
images on both sides of the split	0

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`	`16a44c64a0c2e609`

Provenance and verification


records	5,097
carrying a geometry block	5,097 / 5,097
instances per record	`1`: 203, `2`: 192, `3`: 142, `4`: 139, `5+`: 4,421
total instances	151,428
image dimensions	1920×1080 (1,655), 1280×720 (1,460), 640×480 (805)
`scale` values present	[1.0]

Computed from this repo's own 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)	640×480 → 644×476, 1280×720 → 1288×728, 1280×800 → 1288×812
shipped boxes	151,428
legible at that render (>=16px there)	128,009 (84.5%)

⚠ 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/common/adapt_engine.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 640×480 is rendered 644×476 and native-pixel boxes are then wrong by a few pixels. forge_model/common/adapt_engine.py regenerates coordinates for a target render size, re-derives counts, and drops records whose gold no longer holds there.

Downloads last month: 87

Total file size:

709 MB

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