Saliency e-commerce images (SalECI)

我们建立了新的SalECI数据集，包括972具有收集的固定和注释的文本边界的电子商务图像。从主流平台收集的所有图片，包括淘宝，亚马逊和Wish。图2显示了SalECI中的13个类别及其对应的图像编号。总而言之，我们的SalECI数据集包括来自25名受试者的眼动追踪实验的257，302个注视，以及由3名志愿者注释的10个833个文本边界框。关于数据集建立的细节在补充条款中介绍We developed a novel SalECI dataset, which comprises 972 e-commerce images with collected and annotated text bounding boxes. All images were sourced from mainstream e-commerce platforms including Taobao, Amazon, and Wish. Figure 2 illustrates the 13 categories included in SalECI and their corresponding image counts. Collectively, the SalECI dataset contains 257,302 fixation points derived from eye-tracking experiments involving 25 participants, alongside 10,833 text bounding boxes annotated by three volunteers. Detailed information regarding the construction of this dataset is provided in the supplementary materials.

搜集汇总

背景与挑战

背景概述

SalECI数据集是一个包含972张电子商务图像的数据集，这些图像来自淘宝、亚马逊和Wish等主流平台，并附有25名受试者眼动追踪实验产生的257,302个注视点，以及3名志愿者标注的10,833个文本边界框。该数据集由北京航空航天大学于2022年发布，主要用于研究文本在电商图像中的显著性。