1P1-U02 物体の形状概念と共起を利用した室内環境における一般物体認識(ロボットビジョン)

抄録

In previous, we proposed a general object recognition method to use the general shape concept of object, an example of which for a chair is "it has a seating surface at certain height", "it may have a back surface", etc. This method doesn't need learning some features of object. However, there is a problem that recognition is not possible if the seat and back are apart in such a case of pipe chair. To address this problem, we propose a method to use the general concepts based on not only object shape but also co-occurrence, such as "desk and chair exist in the neighborhood of each other". The effectiveness of proposed method was experimentally verified in practical several situations.