VIA

VIA 数据集是用于帮助视障人士的数据集。提出的数据集 1 由 342 个图像组成，分为两类：其中 175 个是“清晰路径”，167 个是“非清晰”路径。它们是使用智能手机相机拍摄的，并调整为 750 × 1000 像素。智能手机被放置在用户胸部高度并与地面倾斜约 30 到 60 度，因此它可以捕捉到前方几米的路径，并且超出了普通白色手杖的范围

The VIA Dataset is a dataset developed to assist visually impaired individuals. The proposed Dataset 1 consists of 342 images divided into two categories: 175 are "clear path" and 167 are "non-clear path". These images were captured using smartphone cameras and resized to 750 × 1000 pixels. The smartphones were placed at the user's chest height and tilted at an angle of approximately 30 to 60 degrees relative to the ground, so that they can capture the path several meters ahead, which exceeds the range of a standard white cane.

搜集汇总

背景与挑战

背景概述

VIA数据集专为辅助视障人士设计，包含342张智能手机拍摄的图像，分为'清晰路径'和'非清晰路径'两类，图像尺寸为750×1000像素，拍摄时设备置于胸部高度并倾斜30-60度以捕获前方路径。该数据集由Universidade Estadual Paulista于2020年发布，相关资源可通过GitHub和arXiv获取。