The present invention relates to a three-dimensional (3D) image output system for a portable terminal using region-convolution neural network (R-CNN) algorithm-based object recognition, capable of easily and quickly realizing a 3D image required for virtual reality (VR) or augmented reality (AR) in a portable terminal with low system performance, and an output method thereof. According to the present invention, the 3D image output system for a portable terminal using a R-CNN algorithm-based object recognition comprises: the portable terminal storing and transmitting an image or moving image file, which is to be output as a 3D image, photographing an object to be detected to an object recognition server, and extracting a recognition result from the object recognition server to an integrated image for the object matched with a corresponding resu the object recognition server using a reason proposal network (RPN) to specify an object region from the image included in the image or moving image file transmitted from the portable terminal, and using an object recognition algorithm in the specified region to store a result recognizing the object included in the image; and a folio case having a lenticular filter embedded therein to display the integrated image of the corresponding object outputted by the recognition result transmitted from the object recognition server as a 3D image.
展开▼