针织布瑕疵检测

识别自动验布机下经过的针织布料表面缺陷。.
精度对应场景或者限制条件
1）场景限定：只支持针织布料。
2）图片分辨率：8192*1000。
3）设备/镜头：需使用指定品牌的先扫面相机，至少需要8k分辨率。
4) 视野大小：单个8k相机视野建议为1000mm幅宽。
5）检测的种类：粗节、纱痕、纱结、漏针、烂针、断拉架、破洞、大破洞、驳口。
6）最高运行速度40米每分钟。
7）布匹最大宽幅2米。

This dataset is designed for detecting surface defects on knitted fabrics passing through an automatic fabric inspection machine. The detection accuracy is subject to the following applicable scenarios and constraint conditions:
1. Fabric type restriction: Only knitted fabrics are supported.
2. Image resolution requirement: 8192 × 1000 pixels.
3. Equipment and lens requirement: A specified-brand line scan camera with a minimum resolution of 8K is required.
4. Field of view recommendation: The recommended working width for a single 8K camera is 1000 mm.
5. Detectable defect categories: Neps, yarn marks, yarn knots, missed stitches, broken needles, broken Lycra strands, holes, large holes, and seam joints.
6. Maximum operating speed: 40 meters per minute.
7. Maximum fabric width: 2 meters.

搜集汇总

背景与挑战

背景概述

该数据集是一个针织布瑕疵检测算法模型，专注于识别针织布料表面的多种缺陷，如粗节、纱痕、破洞等。使用该数据集需要满足特定的设备要求，包括8k分辨率的相机和1000mm幅宽的视野。