Towel Corner Dataset

该数据集包含了使用机器人装置收集的、具有不同颜色和不同排列的毛巾真实图像，每个毛巾的角落都附有标签。该数据集主要用于训练神经网络，以预测毛巾角落的位置，进而应用于启发式算法中，优化毛巾的平滑处理和折叠过程。该数据集规模为3640张图像，其任务是对毛巾角落进行检测与操作。This dataset includes real-world images of towels with diverse colors and arrangements, collected via a robotic setup, where each towel's corners are annotated with labels. This dataset is primarily intended for training neural networks to predict the positions of towel corners, which are then applied in heuristic algorithms to optimize the smoothing and folding processes of towels. Comprising 3640 images in total, this dataset targets the task of detecting and manipulating towel corners.

搜集汇总

背景与挑战

背景概述

Towel Corner Dataset是LUV框架下的一个数据集，专门用于训练神经网络检测毛巾角落等特征。它利用紫外线荧光涂料实现无监督自动标注，无需人工干预，标注效率极高，成本低廉，并已在实际机器人操作如毛巾平滑和折叠中验证了有效性。