Motif-Sikka-ROI: A Curated and Balanced Image Dataset of Traditional Sikka Ikat Weaving Motifs for Classification and Textile Pattern Analysis

Motif-Sikka-ROI is a curated and balanced image dataset documenting 24 classes of traditional ikat weaving motifs from Sikka Regency, East Nusa Tenggara (NTT), Indonesia. The Sikka ikat weaving tradition is a significant part of the cultural identity of the Sikka people, with each motif carrying distinct symbolic and aesthetic meaning passed down through generations.
Images were collected through direct field photography at multiple locations, including weavers' homes, fabric shops, and cultural exhibitions in Sikka Regency, using smartphones and digital cameras. This multi-device, multi-location approach introduces natural variation in lighting, angle, and background conditions, making the dataset representative of real-world collection scenarios.
Each image was processed using the Roboflow platform, where the primary motif area was manually cropped using Region of Interest (ROI) extraction to eliminate irrelevant background and focus on the core weaving pattern. A strict manual filtering process was applied after ROI extraction, retaining only images with sufficient visual clarity and pattern readability. All images were standardized to 224x224 pixels in PNG format— a resolution widely adopted in deep learning pipelines.
The dataset contains exactly 160 images per class across all 24 classes, resulting in a total of 3,840 images. This balanced composition ensures that no single class dominates during model training, making the dataset suitable for the fair benchmarking of classification algorithms. The dataset is published in an unsplit form to maximize flexibility for downstream research tasks, including motif classification, feature extraction, image enhancement, and cultural heritage digitization.

Motif-Sikka-ROI 是一个经精心整理且类别均衡的图像数据集，记录了印度尼西亚东努沙登加拉省（East Nusa Tenggara，NTT）锡卡摄政区（Sikka Regency）的24类传统伊卡特（ikat）编织纹样。锡卡伊卡特编织传统是锡卡族文化身份的重要组成部分，每一种纹样都承载着世代传承的独特象征意义与美学内涵。
图像采集工作通过实地直接拍摄完成，采集点位涵盖锡卡摄政区内的织工居所、面料商铺与文化展会，拍摄设备包含智能手机与数码相机。这种多设备、多点位的采集方案会带来光照条件、拍摄角度与背景环境的自然差异，令该数据集能够贴合现实采集场景的真实特征。
所有图像均通过Roboflow平台进行预处理：首先采用感兴趣区域（Region of Interest）提取技术，手动裁剪出图像中的核心纹样区域，以剔除无关背景、聚焦核心编织图案。在完成ROI提取后，研究团队执行了严格的手动筛选流程，仅保留视觉清晰度达标、纹样辨识度优异的图像。所有图像均被统一调整为224×224像素的PNG格式——这一分辨率在深度学习流水线中被广泛应用。
该数据集的24个类别均各含160张图像，总计3840张样本。这种均衡的类别分布能够确保模型训练过程中不会出现单一类别占优的情况，使得该数据集适用于分类算法的公平性能基准测试。
该数据集以未拆分的形式发布，以最大化下游研究任务的适配灵活性，可应用于纹样分类、特征提取、图像增强与文化遗产数字化等研究方向。