Authors:
Ping Kuang
and
Haoshuang Wang
Affiliation:
University Of Electronic Science And Technology Of China, China
Keyword(s):
Generative Adversarial Networks, Deep Learning, 3D-GAN.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Computational Intelligence
;
Evolutionary Computing
;
Knowledge Discovery and Information Retrieval
;
Knowledge-Based Systems
;
Machine Learning
;
Soft Computing
;
Symbolic Systems
Abstract:
Recently, Generative Adversarial Networks (GANs) gradually applied to the generation of 3D objects and has achieved remarkable success, but at the same time, it also faces some problems, such as the training instability, low-quality samples and mode collapse. We propose a novel framework, namely 3D Bounding Box Generative Adversarial Network(3D-BBGAN), which can reduce the probability space of generation by adding conditional information. According this way, we can get 3D objects with more detailed geometries.