fabric defect detection

数据集 A：小品类数量布料缺陷数据集共包含4个缺陷类别的2697张图片，包括489张有孔缺陷的织物图像，1300张有线划痕的织物图像，348张有污渍缺陷的织物图像，560张有线缺陷的织物图像。数据集 B：多类别编号织物缺陷数据集包含 21 个缺陷类别，共计 4774 张图像。涉及的缺陷类型有无缺陷、孔洞、污渍、三丝、打结、花板跳、百尺、毛纹、粗经、松经、断经、吊经、粗尺寸、纬缩、膏斑、全经结、星跳、跳花、氨纶断裂、色差锉、痕迹、纱织瑕疵。

Dataset A: The small-category fabric defect dataset consists of 2697 images across 4 defect categories, including 489 fabric images with hole defects, 1300 fabric images with linear scratch defects, 348 fabric images with stain defects, and 560 fabric images with thread defects.
Dataset B: The multi-category numbered fabric defect dataset contains 4774 images spanning 21 defect categories. The involved defect types are no defect, hole, stain, three-type foreign fibers, knot, pattern jump, hundred feet, hairy streak, thick warp yarn, loose warp yarn, broken warp yarn, sagged warp yarn, abnormal dimension, weft shrinkage, paste stain, full warp knot, star skip, pattern skip, spandex breakage, color difference defect, trace defect, and yarn weaving defects.

搜集汇总

背景与挑战

背景概述

该数据集是一个用于布料缺陷检测的图像数据集，包含两个子集：数据集A有4个缺陷类别共2697张图片，数据集B有21个缺陷类别共4774张图像。总规模为7471张图像，覆盖了孔洞、污渍、线缺陷等多种常见织物缺陷类型，适用于训练和评估缺陷检测模型。