ロボットによる衣服操作のためのMask R-CNNを用いた複数衣服の認識

抄録

In the super-ageing society, automating washing, which is a heavy burden among housekeeping, is useful for supporting the lives of non-healthy elderly people. Among washing automation, the accuracy of clothing detection on an image obtained from a robot camera greatly influences clothing operation including griping and carrying of clothes afterwards. In this research, we aim to correctly recognize each garment from multiple clothes. We performed instance segmentation of images containing multiple clothes by Mask R - CNN which is a latest Deep Learning method.