Optical illusion hidden picture is an interesting visual perceptual phenomenon where an image is cleverly integrated into another picture. Established on the off-the-shelf text-to-image (T2I) diffusion model, we propose a novel text-guided image-to-image (I2I) translation framework dubbed as Phase-Transferred Diffusion Model (PTDiffusion) for hidden art syntheses, which harmoniously embeds an input reference image into arbitrary scenes described by the text prompts. At the heart of our method is a plug-and-play phase transfer mechanism that dynamically and progressively transplants diffusion features’ phase spectrum from the denoising process to reconstruct the reference image into the one to sample the generated illusion image, realizing deep fusion of the reference structural information and the textual semantic information. Furthermore, we propose asynchronous phase transfer to enable flexible control over the degree of hidden content discernability. Our method bypasses any model training and fine-tuning process, all while substantially outperforming related methods in image quality, text fidelity, visual discernibility, and contextual naturalness for illusion picture synthesis, as demonstrated by extensive qualitative and quantitative experiments. Our project is publically available at this web page.

Recently, text-to-image diffusion models have emerged as a powerful tool for image-to-image translation (I2I), allowing flexible image translation via user-provided text prompts. This paper proposes frequency-controlled diffusion model (FCDiffusion), an end-to-end diffusion-based framework contributing a novel solution to text-guided I2I from a frequency-domain perspective. At the heart of our framework is a feature-space frequency-domain filtering module based on Discrete Cosine Transform, which extracts image features carrying different DCT spectral bands to control the text-to-image generation process of the Latent Diffusion Model, realizing versatile I2I applications including style-guided content creation, image semantic manipulation, image scene translation, and image style translation. Different from related methods, FCDiffusion establishes a unified text-driven I2I framework suiting diverse I2I application scenarios simply by switching among different frequency control branches. The effectiveness and superiority of our method for text-guided I2I are demonstrated with extensive experiments both qualitatively and quantitatively. Our project is publicly available at: https://xianggao1102.github.io/FCDiffusion/.

Generative 3D part assembly involves understanding part relationships and predicting their 6-DoF poses for assembling a realistic 3D shape. Prior work often focus on the geometry of individual parts, neglecting part-whole hierarchies of objects. Leveraging two key observations: 1) super-part poses provide strong hints about part poses, and 2) predicting super-part poses is easier due to fewer super-parts, we propose a part-whole-hierarchy message passing network for efficient 3D part assembly. We first introduce super-parts by grouping geometrically similar parts without any semantic labels. Then we employ a part-whole hierarchical encoder, wherein a super-part encoder predicts latent super-part poses based on input parts. Subsequently, we transform the point cloud using the latent poses, feeding it to the part encoder for aggregating super-part information and reasoning about part relationships to predict all part poses. In training, only ground-truth part poses are required. During inference, the predicted latent poses of super-parts enhance interpretability. Experimental results on the PartNet dataset show that our method achieves state-of-the-art performance in part and connectivity accuracy and enables an interpretable hierarchical part assembly. Code is available at https://github.com//pkudba/3DHPA.

The convenience of 3D sensors has led to an increase in the use of 3D point clouds in various applications. However, the differences in acquisition devices or scenarios lead to divergence in the data distribution of point clouds, which requires good generalization of point cloud representation learning methods. While most previous methods rely on domain adaptation, which involves fine-tuning pre-trained models on target domain data, this may not always be feasible in real-world scenarios where target domain data may be unavailable. To address this issue, we propose InvariantOODG, which learns invariability between point clouds with different distributions using a two-branch network to extract local-to-global features from original and augmented point clouds. Specifically, to enhance local feature learning of point clouds, we define a set of learnable anchor points that locate the most useful local regions and two types of transformations to augment the input point clouds. The experimental results demonstrate the effectiveness of the proposed model on 3D domain generalization benchmarks.

Large-scale text-to-image diffusion models have been a revolutionary milestone in the evolution of generative AI, allowing wonderful image generation with natural-language text prompt. However, the issue of lacking controllability of such models restricts their practical applicability for real-life content creation. Thus, attention has been focused on leveraging a reference image to control text-to-image synthesis, which is also regarded as manipulating (or editing) a reference image as per a text prompt, namely, text-driven image-to-image translation. This paper contributes a novel, concise, and efficient approach that adapts pre-trained large-scale text-to-image (T2I) diffusion model to the image-to-image (I2I) paradigm in a plug-and-play manner, realizing high-quality and versatile text-driven I2I translation without model training, fine-tuning, or online optimization process. To guide T2I generation with a reference image, we propose to decompose diverse guiding factors with different frequency bands of diffusion features in the DCT spectral space, and accordingly devise a novel frequency band substitution layer which realizes dynamic control of the reference image to the T2I generation result in a plug-and-play manner. We demonstrate that our method allows flexible control over both guiding factor and guiding intensity of the reference image simply by tuning the type and bandwidth of the substituted frequency band, respectively. Extensive qualitative and quantitative experiments verify superiority of our approach over related methods in I2I translation visual quality, versatility, and controllability. Our project is publicly available at: https://xianggao1102.github.io/FBSDiff_webpage/.

3D object representation learning is a fundamental challenge in computer vision to infer about the 3D world. Recent advances in deep learning have shown their efficiency in 3D object recognition, among which view-based methods have performed best so far. However, feature learning of multiple views in existing methods is mostly performed in a supervised fashion, which often requires a large amount of data labels with high costs. In contrast, self-supervised learning aims to learn multi-view feature representations without involving labeled data. To this end, we propose a novel self-supervised framework to learn Multi-View Transformation Equivariant Representations (MV-TER), exploring the equivariant transformations of a 3D object and its projected multiple views that we derive. Specifically, we perform a 3D transformation on a 3D object and obtain multiple views before and after the transformation via projection. Then, we train a representation encoding module to capture the intrinsic 3D object representation by decoding 3D transformation parameters from the fused feature representations of multiple views before and after the transformation. Experimental results demonstrate that the proposed MV-TER significantly outperforms the state-of-the-art view-based approaches in 3D object classification and retrieval tasks and show the generalization to real-world datasets. The code is available at https://github.com/gyshgx868/mvter .

Point clouds are commonly used in a variety of applications such as telepresence, robotics, autonomous driving, etc. However, point clouds are often corrupted by noise, which hinders the performance of point cloud analysis such as segmentation. In this paper, we present a novel approach for robust segmentation of noisy point clouds over graphs, leveraging on graph signal processing. To handle the noise in the input point cloud, we design a feature denoising module that removes the noise from the feature representations, which benefits downstream tasks with the denoised features. In addition, we propose an end-to-end framework that integrates denoising and segmentation tasks for input noisy point clouds, which are jointly optimized by minimizing the segmentation loss and the proposed graph-based denoising loss. We evaluate our method on noisy point clouds and demonstrate the superiority and robustness of the proposed method.

We present the Topology Transformation Equivariant Representation learning, a general paradigm of self-supervised learning for node representations of graph data to enable the wide applicability of Graph Convolutional Neural Networks (GCNNs). We formalize the proposed model from an information-theoretic perspective, by maximizing the mutual information between topology transformations and node representations before and after the transformations. We derive that maximizing such mutual information can be relaxed to minimizing the cross entropy between the applied topology transformation and its estimation from node representations. In particular, we seek to sample a subset of node pairs from the original graph and flip the edge connectivity between each pair to transform the graph topology. Then, we self-train a representation encoder to learn node representations by reconstructing the topology transformations from the feature representations of the original and transformed graphs. In experiments, we apply the proposed model to the downstream node classification, graph classification and link prediction tasks, and results show that the proposed method outperforms the state-of-the-art unsupervised approaches.

Graph Convolutional Neural Networks (GCNNs) extend CNNs to irregular graph data domain, such as brain networks, citation networks and 3D point clouds. It is critical to identify an appropriate graph for basic operations in GCNNs. Existing methods often manually construct or learn one fixed graph based on known connectivities, which may be sub-optimal. To this end, we propose a residual graph learning paradigm to infer edge connectivities and weights in graphs, which is cast as distance metric learning under a low-rank assumption and a similarity-preserving regularization. In particular, we learn the underlying graph based on similarity-preserving mapping on graphs, which keeps similar nodes close and pushes dissimilar nodes away. Extensive experiments on semi-supervised learning of citation networks and 3D point clouds show that we achieve the state-of-the-art performance in terms of both accuracy and robustness.

Point clouds have attracted increasing attention. Significant progress has been made in methods for point cloud analysis, which often requires costly human annotation as supervision. To address this issue, we propose a novel self-contrastive learning for self-supervised point cloud representation learning, aiming to capture both local geometric patterns and nonlocal semantic primitives based on the nonlocal self-similarity of point clouds. The contributions are two-fold: on the one hand, instead of contrasting among different point clouds as commonly employed in contrastive learning, we exploit self-similar point cloud patches within a single point cloud as positive samples and otherwise negative ones to facilitate the task of contrastive learning. On the other hand, we actively learn hard negative samples that are close to positive samples for discriminative feature learning, which are sampled conditional on each anchor patch leveraging on the degree of self-similarity. Experimental results show that the proposed method achieves state-of-the-art performance on widely used benchmark datasets for self-supervised point cloud segmentation and transfer learning for classification.

3D dynamic point clouds provide a natural discrete representation of real-world objects or scenes in motion, with a wide range of applications in immersive telepresence, autonomous driving, surveillance, etc. Nevertheless, dynamic point clouds are often perturbed by noise due to hardware, software or other causes. While a plethora of methods have been proposed for static point cloud denoising, few efforts are made for the denoising of dynamic point clouds, which is quite challenging due to the irregular sampling patterns both spatially and temporally. In this paper, we represent dynamic point clouds naturally on spatial-temporal graphs, and exploit the temporal consistency with respect to the underlying surface (manifold). In particular, we define a manifold-to-manifold distance and its discrete counterpart on graphs to measure the variation-based intrinsic distance between surface patches in the temporal domain, provided that graph operators are discrete counterparts of functionals on Riemannian manifolds. Then, we construct the spatial-temporal graph connectivity between corresponding surface patches based on the temporal distance and between points in adjacent patches in the spatial domain. Leveraging the initial graph representation, we formulate dynamic point cloud denoising as the joint optimization of the desired point cloud and underlying graph representation, regularized by both spatial smoothness and temporal consistency. We reformulate the optimization and present an efficient algorithm. Experimental results show that the proposed method significantly outperforms independent denoising of each frame from state-of-the-art static point cloud denoising approaches, on both Gaussian noise and simulated LiDAR noise.

Recent advances in Graph Convolutional Neural Networks (GCNNs) have shown their efficiency for nonEuclidean data on graphs, which often require a large amount of labeled data with high cost. It it thus critical to learn graph feature representations in an unsupervised manner in practice. To this end, we propose a novel unsupervised learning of Graph Transformation Equivariant Representations (GraphTER), aiming to capture intrinsic patterns of graph structure under both global and local transformations. Specifically, we allow to sample different groups of nodes from a graph and then transform them node-wise isotropically or anisotropically. Then, we self-train a representation encoder to capture the graph structures by reconstructing these node-wise transformations from the feature representations of the original and transformed graphs. In experiments, we apply the learned GraphTER to graphs of 3D point cloud data, and results on point cloud segmentation/classification show that GraphTER significantly outperforms state-of-the-art unsupervised approaches and pushes greatly closer towards the upper bound set by the fully supervised counterparts. The code is available at: https://github.com/gyshgx868/graph-ter.

Graph Convolutional Neural Networks (GCNNs) are generalizations of CNNs to graph-structured data, in which convolution is guided by the graph topology. In many cases where graphs are unavailable, existing methods manually construct graphs or learn task-driven adaptive graphs. In this paper, we propose Graph Learning Neural Networks (GLNNs), which exploit the optimization of graphs (the adjacency matrix in particular) from both data and tasks. Leveraging on spectral graph theory, we propose the objective of graph learning from a sparsity constraint, properties of a valid adjacency matrix as well as a graph Laplacian regularizer via maximum a posteriori estimation. The optimization objective is then integrated into the loss function of the GCNN, which adapts the graph topology to not only labels of a specific task but also the input data. Experimental results show that our proposed GLNN significantly outperforms state-of-the-art approaches over widely adopted social network datasets and citation network datasets for semi-supervised classification.

With the prevalence of accessible depth sensors, dynamic human body skeletons have attracted much attention as a robust modality for action recognition. Previous methods model skeletons based on RNN or CNN, which has limited expressive power for irregular skeleton joints. While graph convolutional networks (GCN) have been proposed to address irregular graph-structured data, the fundamental graph construction remains challenging. In this paper, we represent skeletons naturally on graphs, and propose a graph regression based GCN (GR-GCN) for skeleton-based action recognition, aiming to capture the spatio-temporal variation in the data. As the graph representation is crucial to graph convolution, we first propose graph regression to statistically learn the underlying graph from multiple observations. In particular, we provide spatio-temporal modeling of skeletons and pose an optimization problem on the graph structure over consecutive frames, which enforces the sparsity of the underlying graph for efficient representation. The optimized graph not only connects each joint to its neighboring joints in the same frame strongly or weakly, but also links with relevant joints in the previous and subsequent frames. We then feed the optimized graph into the GCN along with the coordinates of the skeleton sequence for feature learning, where we deploy high-order and fast Chebyshev approximation of spectral graph convolution. Further, we provide analysis of the variation characterization by the Chebyshev approximation. Experimental results validate the effectiveness of the proposed graph regression and show that the proposed GR-GCN achieves the state-of-the-art performance on the widely used NTU RGB+D, UT-Kinect and SYSU 3D datasets.