11 September 2023 in IEEE Transactions on Pattern Analysis and Machine Intelligence
After decades of investigation, point cloud registration is still a challenging task in practice, especially when the correspondences are contaminated by a large number of outliers. It may result in a rapidly decreasing probability of generating a hypothesis close to the true transformation, leading to the failure of point cloud registration. To tackle this problem, we propose a transformation estimation method, named Hunter, for robust point cloud registration with severe outliers. The core of Hunter is to design a global-to-local exploration scheme to robustly find the correct correspondences. The global exploration aims to exploit guided sampling to generate promising initial alignments. To this end, a hypergraph-based consistency reasoning module is introduced to learn the high-order consistency among correct correspondences, which is able to yield a more distinct inlier cluster that facilitates the generation of all-inlier hypotheses. Moreover, we propose a preference-based local exploration module that exploits the preference information of top- $k$ promising hypotheses to find a better transformation. This module can efficiently obtain multiple reliable transformation hypotheses by using a multi-initialization searching strategy. Finally, we present a distance-angle based hypothesis selection criterion to choose the most reliable transformation, which can avoid selecting symmetrically aligned false transformations. Experimental results on simulated, indoor, and outdoor datasets, demonstrate that Hunter can achieve significant superiority over the state-of-the-art methods, including both learning-based and traditional methods (as shown in Fig. 1). Moreover, experimental results also indicate that Hunter can achieve more stable performance compared with all other methods with severe outliers.
19 March 2024 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Self-supervised representation learning for 3D point clouds has attracted increasing attention. However, existing methods in the field of 3D computer vision generally use fixed embeddings to represent the latent features, and impose hard constraints on the embeddings to make the latent feature values of the positive samples converge to consistency, which limits the ability of feature extractors to generalize over different data domains. To address this issue, we propose a Generative Variational-Contrastive Learning (GVC) model, where Gaussian distribution is used to construct a continuous, smoothed representation of the latent features. A distribution constraint and cross-supervision are constructed to improve the transfer ability of the feature extractor over synthetic and real-world data. Specifically, we design a variational contrastive module to constrain the feature distribution instead of feature values corresponding to each sample in the latent space. Moreover, a generative cross-supervision module is introduced to preserve the invariance features and promote the consistency of feature distribution among positive samples. Experimental results demonstrate that GVC achieves SOTA on different downstream tasks. In particular, with only pre-training on the synthetic dataset, GVC achieves a lead of 8.4% and 14.2% when transferring to the real-world dataset in the linear classification and few-shot classification.
29 July 2023 in Applied Intelligence
23 September 2025 in IEEE Transactions on Pattern Analysis and Machine Intelligence
The extraction of road network is essential for the generation of high-definition maps since it enables the precise localization of road landmarks and their interconnections. However, generating road network poses a significant challenge due to the conflicting underlying combination of Euclidean (e.g., road landmarks location) and non-Euclidean (e.g., road topological connectivity) structures. Existing methods struggle to merge the two types of data domains effectively, but few of them address it properly. Instead, our work establishes a unified representation of both types of data domain by projecting both Euclidean and non-Euclidean data into an integer series called RoadNet Sequence. Further than modeling an auto-regressive sequence-to-sequence Transformer model to understand RoadNet Sequence, we decouple the dependency of RoadNet Sequence into a mixture of auto-regressive and non-autoregressive dependency. Building on this, our proposed non-autoregressive sequence-to-sequence approach leverages non-autoregressive dependencies while fixing the gap towards auto-regressive dependencies, resulting in success in both efficiency and accuracy. We further identify two main bottlenecks in the current RoadNetTransformer on a non-overfitting split of the dataset: poor landmark detection limited by the BEV Encoder and error propagation to topology reasoning. Therefore, we propose Topology-Inherited Training to inherit better topology knowledge into RoadNetTransformer. Additionally, we collect SD-Maps from open-source map datasets and use this prior information to significantly improve landmark detection and reachability. Extensive experiments on the nuScenes dataset demonstrate the superiority of RoadNet Sequence representation and the non-autoregressive approach compared to existing state-of-the-art alternatives.
30 November 2023 in IEEE Transactions on Multimedia
Keypoint detection and descriptor matching are two vital steps in the 3D feature extraction framework, but they are difficult to learn in an end-to-end fashion due to their inherent discreteness. To tackle the non-differentiable operations, we formulate feature extraction as a decision-making problem: the network is treated as a policy pool that can make probabilistic estimations for keypoint selection and feature matching, supervised by maximizing a reward expectation of actions. In this way, we propose a novel end-to-end training paradigm of 3D feature extraction based on the stochastic policy gradient method, named Reinforced Detectors and Descriptors (RDD). Firstly, we propose a local-to-global probabilistic keypoint selection module that formulates the sampling probabilities of keypoints in a local-and-global mechanism to yield sparse and accurate keypoints. Secondly, we regard feature matching as an optimal transport problem and an efficient Sinkhorn method is leveraged to solve the optimal matching probabilities. In particular, we carefully design a reward function and derive gradients of probabilistic actions, thus overcoming the discreteness and providing reinforced supervision signals. Since our reward function is calculated from sampled keypoints rather than from randomly sampled points as in existing methods, the gap between training and inference is bridged. Experimental results demonstrate that our approach exceeds the quality of state-of-the-art methods and shows strong generalization ability. Remarkably, our approach can achieve significantly higher Registration Recall than other advanced methods when aligning scenes with a small number of keypoints, due to our highly accurate and repeatable detector.