摘要 |
PROBLEM TO BE SOLVED: To efficiently recognize an operation obtained by combining a plurality of operations. SOLUTION: An object recognition device is configured to generate a learning image showing the shape of a target object being the target of the recognition of an operation, and to accept the input of the image of the target object, and to recognize an operation shown by the image accepted on the basis of the learning image, and provided with: a learning image generation part for generating an image showing a plurality of shapes different from each other in one operation being a series of operations included in the prescribed operation of the target object by operation state of a series of operations other than the first operation included in the prescribed operation, and for setting the generated image as a learning image; and a learning processing part for projecting each of the learning images generated by the learning image generation part as a feature point on a unique space for recognizing the operation shown by the accepted image. COPYRIGHT: (C)2011,JPO&INPIT
|