arXiv subjects

Xu Yan

Publications and source records attributed to Xu Yan.

83 records · Page 5Linked to original sources

arXiv2021

InstanceRefer: Cooperative Holistic Understanding for Visual Grounding on Point Clouds through Instance Multi-level Contextual Referring

Compared with the visual grounding on 2D images, the natural-language-guided 3D object localization on point clouds is more challenging. In this paper, we propose a new model, named InstanceRefer, to achieve a superior 3D visual grounding through the grounding-by-matching strategy. In practice, our model first predicts the target category from the language descriptions using a simple language classification model. Then, based on the category, our model sifts out a small number of instance candidates (usually less than 20) from the panoptic segmentation of point clouds. Thus, the non-trivial 3D visual grounding task has been effectively re-formulated as a simplified instance-matching problem, considering that instance-level candidates are more rational than the redundant 3D object proposals. Subsequently, for each candidate, we perform the multi-level contextual inference, i.e., referring from instance attribute perception, instance-to-instance relation perception, and instance-to-background global localization perception, respectively. Eventually, the most relevant candidate is selected and localized by ranking confidence scores, which are obtained by the cooperative holistic visual-language feature matching. Experiments confirm that our method outperforms previous state-of-the-arts on ScanRefer online benchmark and Nr3D/Sr3D datasets.

cs.CV↗

arXiv2021

PointLIE: Locally Invertible Embedding for Point Cloud Sampling and Recovery

Point Cloud Sampling and Recovery (PCSR) is critical for massive real-time point cloud collection and processing since raw data usually requires large storage and computation. In this paper, we address a fundamental problem in PCSR: How to downsample the dense point cloud with arbitrary scales while preserving the local topology of discarding points in a case-agnostic manner (i.e. without additional storage for point relationship)? We propose a novel Locally Invertible Embedding for point cloud adaptive sampling and recovery (PointLIE). Instead of learning to predict the underlying geometry details in a seemingly plausible manner, PointLIE unifies point cloud sampling and upsampling to one single framework through bi-directional learning. Specifically, PointLIE recursively samples and adjusts neighboring points on each scale. Then it encodes the neighboring offsets of sampled points to a latent space and thus decouples the sampled points and the corresponding local geometric relationship. Once the latent space is determined and that the deep model is optimized, the recovery process could be conducted by passing the recover-pleasing sampled points and a randomly-drawn embedding to the same network through an invertible operation. Such a scheme could guarantee the fidelity of dense point recovery from sampled points. Extensive experiments demonstrate that the proposed PointLIE outperforms state-of-the-arts both quantitatively and qualitatively. Our code is released through https://github.com/zwb0/PointLIE.

cs.CV↗

arXiv2020

Sparse Single Sweep LiDAR Point Cloud Segmentation via Learning Contextual Shape Priors from Scene Completion

LiDAR point cloud analysis is a core task for 3D computer vision, especially for autonomous driving. However, due to the severe sparsity and noise interference in the single sweep LiDAR point cloud, the accurate semantic segmentation is non-trivial to achieve. In this paper, we propose a novel sparse LiDAR point cloud semantic segmentation framework assisted by learned contextual shape priors. In practice, an initial semantic segmentation (SS) of a single sweep point cloud can be achieved by any appealing network and then flows into the semantic scene completion (SSC) module as the input. By merging multiple frames in the LiDAR sequence as supervision, the optimized SSC module has learned the contextual shape priors from sequential LiDAR data, completing the sparse single sweep point cloud to the dense one. Thus, it inherently improves SS optimization through fully end-to-end training. Besides, a Point-Voxel Interaction (PVI) module is proposed to further enhance the knowledge fusion between SS and SSC tasks, i.e., promoting the interaction of incomplete local geometry of point cloud and complete voxel-wise global structure. Furthermore, the auxiliary SSC and PVI modules can be discarded during inference without extra burden for SS. Extensive experiments confirm that our JS3C-Net achieves superior performance on both SemanticKITTI and SemanticPOSS benchmarks, i.e., 4% and 3% improvement correspondingly.

cs.CV↗

arXiv2020

PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling

Raw point clouds data inevitably contains outliers or noise through acquisition from 3D sensors or reconstruction algorithms. In this paper, we present a novel end-to-end network for robust point clouds processing, named PointASNL, which can deal with point clouds with noise effectively. The key component in our approach is the adaptive sampling (AS) module. It first re-weights the neighbors around the initial sampled points from farthest point sampling (FPS), and then adaptively adjusts the sampled points beyond the entire point cloud. Our AS module can not only benefit the feature learning of point clouds, but also ease the biased effect of outliers. To further capture the neighbor and long-range dependencies of the sampled point, we proposed a local-nonlocal (L-NL) module inspired by the nonlocal operation. Such L-NL module enables the learning process insensitive to noise. Extensive experiments verify the robustness and superiority of our approach in point clouds processing tasks regardless of synthesis data, indoor data, and outdoor data with or without noise. Specifically, PointASNL achieves state-of-the-art robust performance for classification and segmentation tasks on all datasets, and significantly outperforms previous methods on real-world outdoor SemanticKITTI dataset with considerate noise. Our code is released through https://github.com/yanx27/PointASNL.

cs.CV↗

arXiv2020

SQLFlow: A Bridge between SQL and Machine Learning

Industrial AI systems are mostly end-to-end machine learning (ML) workflows. A typical recommendation or business intelligence system includes many online micro-services and offline jobs. We describe SQLFlow for developing such workflows efficiently in SQL. SQL enables developers to write short programs focusing on the purpose (what) and ignoring the procedure (how). Previous database systems extended their SQL dialect to support ML. SQLFlow (https://sqlflow.org/sqlflow ) takes another strategy to work as a bridge over various database systems, including MySQL, Apache Hive, and Alibaba MaxCompute, and ML engines like TensorFlow, XGBoost, and scikit-learn. We extended SQL syntax carefully to make the extension working with various SQL dialects. We implement the extension by inventing a collaborative parsing algorithm. SQLFlow is efficient and expressive to a wide variety of ML techniques -- supervised and unsupervised learning; deep networks and tree models; visual model explanation in addition to training and prediction; data processing and feature extraction in addition to ML. SQLFlow compiles a SQL program into a Kubernetes-native workflow for fault-tolerable execution and on-cloud deployment. Current industrial users include Ant Financial, DiDi, and Alibaba Group.

cs.DB↗

arXiv2017

Singlet S-wave superfluidity of proton in neutron star matter

The possible $^{1}S_{0}$ protonic superfluidity is investigated in neutron star matter, and the corresponding energy gap as a function of baryonic density is calculated on the basis of BCS gap equation. We have discussed particularly the influence of hyperon degrees of freedom on $^{1}S_{0}$ protonic superfluidity. It is found that the appearance of hyperons leads to a slight decrease of $^{1}S_{0}$ protonic pairing energy gap in most density range of existing $^{1}S_{0}$ protonic superfluidity. However, when the baryonic density $ρ_{B}>$0.377 (or 0.409) fm${^{-3}}$ for TM1 (or TMA) parameter set, $^{1}S_{0}$ protonic pairing energy gap is significantly larger than the corresponding values without hyperons. And the baryonic density range of existing $^{1}S_{0}$ protonic superfluidity is widen due to the appearance of hyperons. In our results, the hyperons not only change the EOS and bulk properties but also change the size and baryon density range of $^{1}S_{0}$ protonic superfluidity in neutron star matter.

astro-ph.HE↗

arXiv2016

Effects of the tensor couplings on the nucleonic direct URCA processes in neutron star matter

The relativistic neutrino emissivity of the nucleonic direct URCA processes in neutron star matter are investigated within the relativistic Hartree-Fock approximation. We study particularly the influences of the tensor couplings of vector mesons $ω$ and $ρ$ on the nucleonic direct URCA processes. It is found that the inclusion of the tensor couplings of vector mesons $ω$ and $ρ$ can slightly increase the maximum mass of neutron stars. In addition, the results indicate that the tensor couplings of vector mesons $ω$ and $ρ$ lead to obvious enhancement of the total neutrino emissivity for the nucleonic direct URCA processes, which must accelerate the cooling rate of the non-superfluid neutron star matter. However, when considering only the tensor coupling of vector meson $ρ$, the neutrino emissivity for the nucleonic direct URCA processes slightly declines at low densities and significantly increases at high densities.That is to say that the tensor coupling of vector meson $ρ$ leads to the slow cooling rate of a low-mass neutron star and rapid cooling rate of a massive neutron star.

nucl-th↗

arXiv2015

Classifying Relations via Long Short Term Memory Networks along Shortest Dependency Path

Relation classification is an important research arena in the field of natural language processing (NLP). In this paper, we present SDP-LSTM, a novel neural network to classify the relation of two entities in a sentence. Our neural architecture leverages the shortest dependency path (SDP) between two entities; multichannel recurrent neural networks, with long short term memory (LSTM) units, pick up heterogeneous information along the SDP. Our proposed model has several distinct features: (1) The shortest dependency paths retain most relevant information (to relation classification), while eliminating irrelevant words in the sentence. (2) The multichannel LSTM networks allow effective information integration from heterogeneous sources over the dependency paths. (3) A customized dropout strategy regularizes the neural network to alleviate overfitting. We test our model on the SemEval 2010 relation classification task, and achieve an $F_1$-score of 83.7\%, higher than competing methods in the literature.

cs.CL↗

arXiv2014

Superconductivity in a two-dimensional superconductor with Rashba and Dresselhaus spin-orbit couplings

We present a general model with both Rashba and Dresselhaus spin-orbit couplings to describe a two-dimensional noncentrosymmetric superconductor. The combined effects of the two spin-orbit couplings on superconductivity are investigated in the framework of mean-field theory. We find that the Rashba and Dresselhaus spin-orbit couplings result in similar effects on superconductivity if they are present solely in the system. Mixing of spin-singlet and triplet pairings in electron band is induced under the assumption that each quasiparticle band is p-wave paired. If the two types of spin-orbit couplings appear jointly, both the singlet and triplet pairings are weakened and decreased down to their minimum values in the equal-Rashba-Dresselhaus case.

cond-mat.supr-con↗

arXiv2013

Electron correlation and impurity-induced quasiparticle resonance states in cuprate superconductors

We theoretically study the quasiparticle resonance states around a nonmagnetic impurity in cuprate superconductors based on t-t'-J-U model. The purpose of introducing the Coulomb repulsive interaction U is to partially impose the double occupancy constraint by employing the Gutzwiller projected mean-field approximation. We determine the spatial variation of the order parameter and the local density of states (LDOS) by self-consistently solving Bogoliubov-de Gennes equations. We find that in the large U limit, a zero-energy resonance peak in the LDOS indeed appears for the impurity potential in the unitary limit, at the same time the asymmetric superconducting coherence peaks are strongly suppressed. As U decreases the electron double occupancy d is permitted and gradually increases, leading to the decreasing of order parameter. In particular the above zero-energy resonance peak begins to evolve into a double-peaked structure since a critical value $d_{c}$. These important feathers are qualitatively agreement with the scanning tunneling spectroscopy experiments, and uncover the essential role played by the electron correlation in cuprate superconductors.

cond-mat.supr-con↗

arXiv2013

Coexistence of p-wave Cooper pairing and ferromagnetism

A two-band model for coexistence of p-wave superconductivity with localized ferromagnetism is studied using the equation of motion approach. It shows that ferromagnetic and superconducting states enhance each other but in a different way from that of the one-band model. The Curie temperature is not only determined by the exchange interactions between localized spins, but also can be increased with the coupling between electrons and spins, and with the p-wave Cooper-pairing interaction. These results are complementary to those of the one-band model, which suggest that the Curie temperature is unlikely to ever be below the superconducting transition temperature.

cond-mat.supr-con↗