In real-world scenarios, scanned point clouds are often incomplete due to occlusion issues. The tasks of self-supervised and weakly-supervised point cloud completion involve reconstructing missing regions of these incomplete objects without the supervision of complete ground truth. Current methods either rely on multiple views of partial observations for supervision or overlook the intrinsic geometric similarity that can be identified and utilized from the given partial point clouds. In this paper, we propose MAL-UPC, a framework that effectively leverages both region-level and category-specific geometric similarities to complete missing structures. Our MAL-UPC does not require any 3D complete supervision and only necessitates single-view partial observations in the training set. Specifically, we first introduce a Pattern Retrieval Network to retrieve similar position and curvature patterns between the partial input and the predicted shape, then leverage these similarities to densify and refine the reconstructed results. Additionally, we render the reconstructed complete shape into multi-view depth maps and design an adversarial learning module to learn the geometry of the target shape from category-specific single-view depth images of the partial point clouds in the training set. To achieve anisotropic rendering, we design a density-aware radius estimation algorithm to improve the quality of the rendered images. Our MAL-UPC outperforms current state-of-the-art self-supervised methods and even some unpaired approaches.

Accurate 3D scene understanding in outdoor environments heavily relies on high-quality point clouds. However, LiDAR-scanned data often suffer from extreme sparsity, severely hindering downstream 3D perception tasks. Existing point cloud upsampling methods primarily focus on individual objects, thus demonstrating limited generalization capability for complex outdoor scenes. To address this issue, we propose PVNet, a diffusion model-based point-voxel interaction framework to perform LiDAR point cloud upsampling without dense supervision. Specifically, we adopt the classifier-free guidance-based DDPMs to guide the generation, in which we employ a sparse point cloud as the guiding condition and the synthesized point clouds derived from its nearby frames as the input. Moreover, we design a voxel completion module to refine and complete the coarse voxel features for enriching the feature representation. In addition, we propose a point-voxel interaction module to integrate features from both points and voxels, which efficiently improves the environmental perception capability of each upsampled point. To the best of our knowledge, our approach is the first scene-level point cloud upsampling method supporting arbitrary upsampling rates. Extensive experiments on various benchmarks demonstrate that our method achieves state-of-the-art performance. The source code will be available at https://github.com/chengxianjing/PVNet.

The traditional view-aware based vehicle re-identification methods integrated the vehicle view information into the training process, allowing the model parameters to learn the vehicle view knowledge. So, the extracted vehicle features contained vehicle view information, and the impact of view differences on model performance can be reduced. Since ensemble training with different methods will affect each other, it is difficult for these methods to train models together with other non view-aware methods. In order to address this challenge, in this paper, we propose a view-aware-based post-processing method (VABPP), which uses vehicle view information to re-rank the re-identification results during testing. According to vehicle views, VABPP method divide the distances between vehicle id features into several groups. During testing, multiply the distances of each different group by different coefficients. So, it treats different views equally. In order to achieve better implementation of this post-processing method, in this paper, we also propose a scheme that unifies the feature distance distributions of the training set and the test set. This scheme can enable some properties of the training set to be directly used in the test set. As the properties of the training set are trained under the guidance of correctly labeled labels, which enhances the robustness of the test set properties. The mAP in VeRi-776 dataset and in the three test sets of VERI-Wild dataset are 83.7%, 88.7%, 84.6% and 78.5%, respectively. And the rank-1 in the three test sets of VehicleID dataset are 88.6%, 85.4% and 81.1%, respectively.

Point cloud upsampling is an essential yet challenging task in various 3D computer vision and graphics applications. Existing methods often struggle with limitations such as the generation of outliers or shrinkage artifacts. Additionally, these methods usually ignore the overall spatial structure of point clouds, leading to suboptimal results. To tackle these challenges, we propose a novel framework that enhances geometric spatial consistency in upsampled point clouds through a dual‐supervision mechanism and enables the generation of high‐fidelity results with precise geometric structures. Specifically, we first design a tailored feature extractor that iteratively extracts the comprehensive and distinctive features by integrating both fine‐grained local geometric details and global structure information. Then, our network predicts the point‐to‐point distances and Chamfer distances of upsampled points to accurately capture the spatial relation within them. To enhance spatial consistency, we formulate a joint loss function that enables our model to perceive the spatial relations between points by indirect and direct supervision. This ensures the precise alignment between upsampled points and ground truth during training. Furthermore, we propose a coordinate reconstruction to generate more high‐quality upsampled points iteratively. We conduct extensive experiments across multiple benchmark datasets and downstream tasks. The results comprehensively demonstrate that our method achieves state‐of‐the‐art performance and exhibits superior generalisation capabilities.

It is well known that preserving depth edges is an effective solution for achieving the accurate disparity map in stereo matching, but many state‐of‐the‐art methods do not preserve depth edges well. In order to solve it efficiently, the cell structure containing irregular and regular shape regions is designed to preserve depth edges. Based on the well‐designed cell structure, a novel disparity estimation method for stereo matching is proposed, in which a two‐layer disparity optimization method is proposed to refine the disparity plane; it includes the front‐parallel disparities computation and slanted‐surfaces disparity plane refinement. In the framework of front‐parallel disparities computation, a tree‐based cost aggregation method is presented to make full use of the segmentation information of cells and then performing semi‐global cost aggregation. In the framework of slanted‐surfaces disparity plane refinement, a new probability model is proposed that employs Bayesian inference for refining disparities in textureless, weak texture and occluded regions. Experimental results show that higher accuracy could be achieved via the proposed method compared with some known state‐of‐the‐art stereo methods on KITTI 2015 and Middlebury dataset, which are the standard benchmarks for testing the stereo matching methods. It can also be indicated that the proposed method can produce accurate disparity map and have good generalization performance.