Object identification device, mobile system, object identification method, object identification model learning method and object identification model learning device

摘要

This object identification device includes: an image acquisition unit (40a) which acquires an outside image (IM) captured by an image capturing unit from a prescribed image capturing viewpoint (PVP); and viewpoint conversion map generation units (60, 260) in which a convolutional neural network is formed, data of an image is input to the convolutional neural network, and a viewpoint conversion map (MB) by which the image of an image capturing viewpoint is converted into the image of another viewpoint is output via the convolutional neural network. The viewpoint conversion map generation unit includes: feature extraction processing units (61, 261, 266) which extract a feature map of an object in a first coordinate system (CV) and output processing units (65, 66, 271) which output a viewpoint conversion map in which the object is identified.