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

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

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

Log in or Sign Up to review the conditions and access this dataset content.

180-grounding

Detection-format defect localization on woven fabric — 245 items (141 good + 104 defective), derived deterministically from the binary segmentation masks of AI4Manufacturing/180. The model outputs boxes as text; defect-free images must output [] — detection rejection is part of the task.

Task

"Locate every defect." annot is a JSON list of {"type": ..., "bbox_xywh": [x, y, w, h]} in the image's pixel coordinates (origin top-left; see metadata.image_wh), one entry per defect instance (connected components after proximity grouping; sub-15-px groups denoised), sorted (type, x, y); the type is one of Broken end, Broken pick, Broken yarn, Contamination, Crease, Cut selvage, Fuzzyball, Knots, Nep, Warp ball, Weft crack, Weft curling. Good images have annot = [] (141). 9 defective images are multi-instance (≥2 boxes).

Verified: every box list re-derived independently from the mask at build — byte-identical on all 245 rows; goods all []; every box within image bounds.

field meaning
query task prompt + JSON output spec (closed class list)
image the woven fabric photo (no overlays)
annot JSON box list, [] when defect-free
reasoning null — deterministic
cate / task B / T-B2
metadata source, image_sha256, image_path, image_wh, r180_record_id, defect_type, n_instances

Provenance

Built deterministically (no LLM/teacher; reasoning is null) from AI4Manufacturing/180 — AITEX woven-fabric surface inspection — 247 fabric-strip photos, 12 defect types + good; each anomalous image has a paired binary pixel segmentation mask (binarized at gray>40). Generator: annotate/180/build_180_derived.py in forge_model; machine gates annotate/180/verify_180.py (all green at build: boxes/gold re-derived byte-identical, clean regions contain zero defect pixels, mcq gold = the true panel, no leaked-artifact vocabulary, image set sha256-disjoint from all other cached AI4Manufacturing datasets).

Resolution. Images are native 4096×256 grayscale strips (16:1 aspect; not downscaled); all coordinates live in that pixel space (see metadata.image_wh).

Query diversity. query is drawn from a fixed pool of surface variants (paraphrases preserving the task + answer format), selected by an independent per-record hash; a machine gate confirms no template correlates with the gold.

The repository name is an internal task code (the source dataset's code is 180).

Downloads last month
14