Longitudinal Shoe Study: Film and Powder Images

Images of shoe prints from 160 pairs of shoes (“Nike Winflo 4” or “Adidas Seeley”; 4 sizes each) made using fingerprint powder on clear adhesive film (contact paper), with associated measurements and data about shoe wear, surfaces, and wearers. Each pair of shoes was worn for at least 10,000 steps per week over a 6-month period, with multiple measurements of the shoe soles taken initially and during three check-in periods spaced at approximately 5 week intervals. The images are accompanied by 3 CSV files describing the shoes, visit (information collected from surveys along with the shoes), and individual images. The codebooks contain descriptions of the variables in each of the CSV files as well as a more extensive description of the file naming scheme outlined in the README. These files can be used to examine wear pattern development, to look for the presence of identifying characteristics among shoes with similar features, and to develop algorithms for matching shoes based on individualizing characteristics. The film and powder prints are likely to contain the most detail of any of the collection methods used in this project and are the most suitable for research on randomly acquired characteristics.

本数据集包含160双鞋的鞋印图像，涉及"Nike Winflo 4"与"Adidas Seeley"两款鞋型，各对应4种尺码。鞋印通过指纹粉印制于透明不干胶接触膜（clear adhesive film, contact paper）上，同时附带与鞋履磨损状态、踩踏表面及穿着者相关的配套测量数据。每双鞋每周至少穿着行走10000步，整个实验周期为6个月；在初始阶段及后续三次间隔约5周的回访检测节点，均对鞋底进行了多次测量。该图像集附带3个CSV文件，分别用于记录鞋履基础信息、回访调查数据（结合鞋履采集的问卷信息）以及单张图像的关联信息。配套代码手册不仅对每个CSV文件内的变量进行了详细说明，还在README文档中完整阐述了项目的文件命名规范。本数据集可用于研究鞋履磨损模式的演化过程，探寻相似特征鞋款间的个体识别特征，以及开发基于个性化特征的鞋履匹配算法。相较于本项目采用的其他鞋印采集方式，透明不干胶接触膜与指纹粉印制的鞋印大概率拥有最丰富的细节表现，最适合用于随机获取的个体特征相关研究。