Recognition processing device, recognition processing method, and program

Provided are a recognition processing device, a recognition processing method, and a program capable of efficiently narrowing down three-dimensional regions on which a recognition process using a three-dimensional convolutional neural network is to be performed. A first recognition process execution unit (68) performs a first recognition process on an image that was captured of a real space, and that is used to generate voxel data. On the basis of the results of the first recognition process, a target two-dimensional region determination unit (70) determines a two-dimensional region constituting a portion of the image captured of the real space. A target three-dimensional region determination unit (72) determines a three-dimensional region in the real space on the basis of the two-dimensional region and the position of the camera that captured the image of the real space. A second recognition process execution unit (74) performs a second recognition process, which uses a three-dimensional convolutional neural