Graphical Models最新文献

In traditional deep functional maps for non-rigid shape correspondence, estimating a functional map including high-frequency information requires enough linearly independent features via the least square method, which is prone to be violated in practice, especially at an early stage of training, or costly post-processing, e.g. ZoomOut. In this paper, we propose a novel method called RFMNet (Robust Deep Functional Map Networks), which jointly considers training stability and more geometric shape features than previous works. We directly first produce a pointwise map by resorting to optimal transport and then convert it to an initial functional map. Such a mechanism mitigates the requirements for the descriptor and avoids the training instabilities resulting from the least square solver. Benefitting from the novel strategy, we successfully integrate a state-of-the-art geometric regularization for further optimizing the functional map, which substantially filters the initial functional map. We show our novel computing functional map module brings more stable training even under encoding the functional map with high-frequency information and faster convergence speed. Considering the pointwise and functional maps, an unsupervised loss is presented for penalizing the correspondence distortion of Delta functions between shapes. To catch discretization-resistant and orientation-aware shape features with our network, we utilize DiffusionNet as a feature extractor. Experimental results demonstrate our apparent superiority in correspondence quality and generalization across various shape discretizations and different datasets compared to the state-of-the-art learning methods.

We introduce a neural network, MixNet, for learning implicit representations of 3D subtle models with large smooth areas and exact shape details in the form of interpolation of two different implicit functions. Our network takes a point cloud as input and uses conventional MLP networks and SIREN networks to predict different implicit fields. We use a learnable interpolation function to combine the implicit values of these two networks and achieve the respective advantages of them. The network is self-supervised with only reconstruction loss, leading to faithful 3D reconstructions with smooth planes, correct details, and plausible spatial partition without any ground-truth segmentation. We evaluate our method on ABC, the largest and most diverse CAD dataset, and some typical shapes to test in terms of geometric correctness and surface smoothness to demonstrate superiority over current alternatives suitable for shape reconstruction.

This paper presents a fast and progressive polygonal approximation algorithm for online strokes. A stroke is defined as a sequence of points between a pen-down and a pen-up. The proposed method generates polygonal approximations progressively as the user inputs the stroke. The proposed algorithm is suitable for real time shape modeling and retrieval. The number of operations used in the proposed algorithm is bounded by O(n), where n is the number of points in a stroke. Detailed experimental results show that the proposed method is not only fast, but also accurate enough compared to other reported algorithms.

In this paper, we present an inverse rendering method for the simple reconstruction of shape and appearance of real-world objects from only roughly calibrated RGB images captured under collocated point light illumination. To this end, we gradually reconstruct the lower-frequency geometry information using automatically generated occupancy mask images based on a visual hull initialization of the mesh, to infer the object topology, and a smoothness-preconditioned optimization. By combining this geometry estimation with learning-based SVBRDF parameter inference as well as intrinsic and extrinsic camera parameter refinement in a joint and unified formulation, our novel method is able to reconstruct shape and an isotropic SVBRDF from fewer input images than previous methods. Unlike in other works, we also estimate normal maps as part of the SVBRDF to capture and represent higher-frequency geometric details in a compact way. Furthermore, by regularizing the appearance estimation with a GAN-based SVBRDF generator, we are able to meaningfully limit the solution space. In summary, this leads to a robust automatic reconstruction algorithm for shape and appearance. We evaluated our algorithm on synthetic as well as on real-world data and demonstrate that our method is able to reconstruct complex objects with high-fidelity reflection properties in a robust way, also in the presence of imperfect camera parameter data.

This paper presents a novel approach for text/speech-driven animation of a photo-realistic head model based on blend-shape geometry, dynamic textures, and neural rendering. Training a VAE for geometry and texture yields a parametric model for accurate capturing and realistic synthesis of facial expressions from a latent feature vector. Our animation method is based on a conditional CNN that transforms text or speech into a sequence of animation parameters. In contrast to previous approaches, our animation model learns disentangling/synthesizing different acting-styles in an unsupervised manner, requiring only phonetic labels that describe the content of training sequences. For realistic real-time rendering, we train a U-Net that refines rasterization-based renderings by computing improved pixel colors and a foreground matte. We compare our framework qualitatively/quantitatively against recent methods for head modeling as well as facial animation and evaluate the perceived rendering/animation quality in a user-study, which indicates large improvements compared to state-of-the-art approaches.

To deal with the exhausting annotations, self-supervised representation learning from unlabeled point clouds has drawn much attention, especially centered on augmentation-based contrastive methods. However, specific augmentations hardly produce sufficient transferability to high-level tasks on different datasets. Besides, augmentations on point clouds may also change underlying semantics. To address the issues, we propose a simple but efficient augmentation fusion contrastive learning framework to combine data augmentations in Euclidean space and feature augmentations in feature space. In particular, we propose a data augmentation method based on sampling and graph generation. Meanwhile, we design a data augmentation network to enable a correspondence of representations by maximizing consistency between augmented graph pairs. We further design a feature augmentation network that encourages the model to learn representations invariant to the perturbations using an encoder perturbation. We comprehensively conduct extensive object classification experiments and object part segmentation experiments to validate the transferability of the proposed framework. Experimental results demonstrate that the proposed framework is effective to learn the point cloud representation in a self-supervised manner, and yields state-of-the-art results in the community. The source code is publicly available at: https://github.com/VCG-NJUST/AFSRL.

Time-varying effects simulation plays a critical role in computer graphics. Fruit diseases are typical time-varying phenomena. Due to the biological complexity, the existing methods fail to represent the biodiversity and biological law of symptoms. To this end, this paper proposes a biology-aware, physically-based framework that respects biological knowledge for realistic simulation of fruit mildew diseases. The simulated symptoms include skin discoloration, fungus growth, and volume shrinkage. Specifically, we take advantage of both the zero-order kinetic model and reaction–diffusion model to represent the complex fruit skin discoloration related to skin biological characteristics. To reproduce 3D mildew growth, we employ the Poisson-disk sampling technique and propose a template model instancing method. One can flexibly change hyphal template models to characterize the fungal biological diversity. To model the fruit’s biological structure, we fill the fruit mesh interior with particles in a biologically-based arrangement. Based on this structure, we propose a turgor pressure and a Lennard-Jones force-based adaptive mass–spring system to simulate the fruit shrinkage in a biological manner. Experiments verified that the proposed framework can effectively simulate mildew diseases, including gray mold, powdery mildew, and downy mildew. Our results are visually compelling and close to the ground truth. Both quantitative and qualitative experiments validated the proposed method.

Assembling components of man-made objects to create new structures or complete 3D shapes is a popular approach in 3D modeling techniques. Recently, leveraging deep neural networks for assembly-based 3D modeling has been widely studied. However, exploring new component combinations even across different categories is still challenging for most of the deep-learning-based 3D modeling methods. In this paper, we propose a novel generative autoencoder that tackles the component combinations for 3D modeling of man-made objects. We use the segmented input objects to create component volumes that have redundant components and random configurations. By using the input objects and the associated component volumes to train the autoencoder, we can obtain an object volume consisting of components with proper quality and structure as the network output. Such a generative autoencoder can be applied to either multiple object categories for structure hybrid or a single object category for shape completion. We conduct a series of evaluations and experimental results to demonstrate the usability and practicability of our method.

Real time Path-tracing is becoming an important approach for the future of games, digital entertainment, and virtual reality applications that require realism and immersive environments. Among different possible optimizations, denoising Monte Carlo rendered images is necessary in low sampling densities. When dealing with Virtual Reality devices, other possibilities can also be considered, such as foveated rendering techniques. Hence, this work proposes a novel and promising rendering pipeline for denoising a real-time path-traced application in a dual-screen system such as head-mounted display (HMD) devices. Therefore, we leverage characteristics of the foveal vision by computing G-Buffers with the features of the scene and a buffer with the foveated distribution for both left and right screens. Later, we path trace the image within the coordinates buffer generating only a few initial rays per selected pixel, and reconstruct the noisy image output with a novel non-homogeneous denoiser that accounts for the pixel distribution. Our experiments showed that this proposed rendering pipeline could achieve a speedup factor up to 1.35 compared to one without our optimizations.