SG-GAN Adversarial Self-Attention GCN for Point Cloud Topological Parts Generation
ABSTARCT :
Point clouds are fundamental in the representation of 3D objects. However, they can also be highly unstructured and irregular. This makes it difficult to directly extend 2D generative models to three-dimensional space. In this paper, we cast the problem of point cloud generation as a topological representation learning problem. To infer the representative information of 3D shapes in the latent space, we propose a hierarchical mixture model that integrates self-attention with an inference tree structure for constructing a point cloud generator. Based on this, we design a novel Generative Adversarial Network (GAN) architecture that is capable to generate realistic point clouds in an unsupervised manner. The proposed adversarial framework (SG-GAN) relies on self-attention mechanism and Graph Convolution Network (GCN) to hierarchically infer the latent topology of 3D shapes. Embedding and transferring the global topology information in a tree framework allows our model to capture and enhance the structural connectivity. Furthermore, the proposed architecture endows our model with partially generating 3D structures. Finally, we propose two gradient penalty methods to stabilize the training of SG-GAN and overcome the possible mode collapse of GAN networks. To demonstrate the performance of our model, we present both quantitative and qualitative evaluations and show that SG-GAN is more efficient in training and it exceeds the state-of-the-art in 3D point cloud generation.
EXISTING SYSTEM :
? We evaluate ParaNet over shape classification and point cloud upsampling, in which our solutions perform favorably against the existing state-of-the-art methods.
? Therefore, most of the existing techniques formulate parameterization as an optimization problem, aiming at finding a balance between distortion and parameterization quality.
? In general, existing studies devote to tailoring convolution-like operators and CNNlike architectures for point cloud feature extraction.
? Empirically, we observe performance decrease when the SAM or the CAM modules are removed from the full pipeline, which implies that our method enables direct adaptation of existing visual techniques that have been proven to be effective in image domain applications.
DISADVANTAGE :
? We propose using TTUR specifically to compensate for the problem of slow learning in a regularized discriminator, making it possible to use fewer discriminator steps per generator step.
? Using these filters, they significantly accelerated the spectral decomposition process, which was one of the main computational bottlenecks in traditional graph convolution problems with large datasets.
? Because the aforementioned GCNs were originally designed for classification problems, the connectivity of graphs was assumed to be given as prior knowledge.
? However, this setting is not appropriate for problems of dynamic model generation.
PROPOSED SYSTEM :
• The proposed TreeGCN preserves the ancestor information of each point and utilizes this information to extract new points via graph convolutions.
• The performance of traditional GCNs can be improved significantly by adopting the proposed tree structures for graph convolutions.
• Based on the proposed tree structures, tree-GAN can generate parts of objects by selecting particular ancestors.
• In contrast, the proposed tree-GAN can not only deal with unordered points, but also extract semantic parts of objects.
• The proposed TreeGCN introduces a tree structure for hierarchical GCNs by passing information from ancestors to descendants of vertices.
ADVANTAGE :
? In addition to self-attention, we also incorporate recent insights relating network conditioning to GAN performance.
? We argue that the generator can also benefit from spectral normalization, based on recent evidence that the conditioning of the generator is an important causal factor in GANs’ performance.
? This comparison demonstrates that the performance improvement given by using SAGAN is not simply due to an increase in model depth and capacity.
? Increasing the size of the convolution kernels can increase the representational capacity of the network but doing so also loses the computational and statistical efficiency obtained by using local convolutional structure.
|