Search

Traditional camera 3D object detectors are typically trained to recognize a predefined set of known object classes. In real-world scenarios, these detectors may encounter unknown objects outside the training categories and fail to identify them correctly. To address this gap, we present OS-Det3D (Open-set Camera 3D Object Detection), a two-stage training framework enhancing the ability of camera 3D detectors to identify both known and unknown objects. The framework involves our proposed 3D Object Discovery Network (ODN3D), which is specifically trained using geometric cues such as the location and scale of 3D boxes to discover general 3D objects. ODN3D is trained in a class-agnostic manner, and the provided 3D object region proposals inherently come with data noise. To boost accuracy in identifying unknown objects, we introduce a Joint Objectness Selection (JOS) module. JOS selects the pseudo ground truth for unknown objects from the 3D object region proposals of ODN3D by combining the ODN3D objectness and camera feature attention objectness. Experiments on the nuScenes and KITTI datasets demonstrate the effectiveness of our framework in enabling camera 3D detectors to successfully identify unknown objects while also improving their performance on known objects.

Out-of-distribution (OOD) detection methods have been developed to identify objects that a model has not seen during training. The Outlier Exposure (OE) methods use auxiliary datasets to train OOD detectors directly. However, the collection and learning of representative OOD samples may pose challenges. To tackle these issues, we propose the Outlier Aware Metric Learning (OAML) framework. The main idea of our method is to use the k-NN algorithm and Stable Diffusion model to generate outliers for training at the feature level without making any distributional assumptions. To increase feature discrepancies in the semantic space, we develop a mutual information-based contrastive learning approach for learning from OOD data effectively. Both theoretical and empirical results confirm the effectiveness of this contrastive learning technique. Furthermore, we incorporate knowledge distillation into our learning framework to prevent degradation of in-distribution classification accuracy. The combination of contrastive learning and knowledge distillation algorithms significantly enhances the performance of OOD detection. Experimental results across various datasets show that our method significantly outperforms previous OE methods.

Inspired by recent theoretical proposal on the interesting piezomagnetism and C-paired valley polarization in V2Se2O monolayer, we predict a stable antiferromagnetic Janus monolayer V2SeTeO with altermagnetic configuration using density functional theory calculations. It exhibits a novel multi-piezo effect combining piezoelectric, piezovalley and piezomagnetism. Most interestingly, the valley polarization and the net magnetization under strain in V2SeTeO exceed these in V2Se2O, along with the additional large piezoelectric coefficient of e31 (0.322*10-10 C m-1). The multi-piezo effect makes antiferromagnetic Janus monolayer V2SeTeO a tantalizing material for potential applications in nanoelectronics, optoelectronics, spintronics and valleytronics., Comment: 20 pages,9 figures

LiDAR point cloud segmentation is one of the most fundamental tasks for autonomous driving scene understanding. However, it is difficult for existing models to achieve both high inference speed and accuracy simultaneously. For example, voxel-based methods perform well in accuracy, while Bird's-Eye-View (BEV)-based methods can achieve real-time inference. To overcome this issue, we develop an effective 3D-to-BEV knowledge distillation method that transfers rich knowledge from 3D voxel-based models to BEV-based models. Our framework mainly consists of two modules: the voxel-to-pillar distillation module and the label-weight distillation module. Voxel-to-pillar distillation distills sparse 3D features to BEV features for middle layers to make the BEV-based model aware of more structural and geometric information. Label-weight distillation helps the model pay more attention to regions with more height information. Finally, we conduct experiments on the SemanticKITTI dataset and Paris-Lille-3D. The results on SemanticKITTI show more than 5% improvement on the test set, especially for classes such as motorcycle and person, with more than 15% improvement. The code can be accessed at https://github.com/fengjiang5/Knowledge-Distillation-from-Cylinder3D-to-PolarNet., Comment: ICME 2023 Accepted

Layered transition metal dichalcogenide 1T-NbSe_2 is a good candidate to explore the charge density wave (CDW) and Mott physics. However, the effects of stacking orders and interlayer coupling in CDW 1T-NbSe_2 are still less explored and understood. Using density functional theory calculations, we present a systematic study of the electronic and magnetic properties of monolayer and layered CDW 1T-NbSe_2. Our results indicate that monolayer CDW 1T-NbSe_2 is a magnetic insulator with \sqrt13\times\sqrt13 periodic lattice modulation. Nevertheless, the magnetic properties of bilayer CDWs 1T-NbSe_2 are found stacking orders dependence. The mechanism is understood by the changes of local magnetic moments in each layer due to spin charge transfer between interlayers. Furthermore, the bulk CDW 1T-NbSe_2 opens a band gap with 0.02 eV in 1\times 1 \times 2 supercell due to the interlayer spin coupling. We also discover that the electronic structures of layered 1T-NbSe_2 show a strong dependence on stacking configurations and dimensionality.

Two-dimensional (2D) hybrid organic-inorganic perovskites (HOIPs) are introducing new directions in the 2D materials landscape. The coexistence of ferroelectricity and spin-orbit interactions play a key role in their optoelectronic properties. We perform a detailed study on a recently synthesized ferroelectric 2D-HOIP, (AMP)PbI$_4$ (AMP = 4-aminomethyl-piperidinium). The calculated polarization and Rashba parameter are in excellent agreement with experimental values. We report a striking new effect, i.e., an extraordinarily large Rashba anisotropy that is tunable by ferroelectric polarization: as polarization is reversed, not only the spin texture chirality is inverted, but also the major and minor axes of the Rashba anisotropy ellipse in k-space are interchanged - a pseudo rotation. A $k \cdot p$ model Hamiltonian and symmetry-mode analysis reveal a quadrilinear coupling between the cation-rotation modes responsible for the Rashba ellipse pseudo-rotation, the framework rotation, and the polarization. These findings may provide new avenues for spin-optoelectronic devices such as spin valves or spin FETs., Comment: 19 pages, 5 figures

Topological Dirac semimetals are a class of semimetals that host symmetry-protected Dirac points near the Fermi level, which arise due to a band inversion of the conduction and valence bands. In this work, we study the less explored class of \emph{noncentrosymmetric} topological Dirac semimetals in three dimensions. We identify the noncentrosymmetric crystallographic point groups required to stabilize fourfold degenerate band crossings and derive model Hamiltonians for all distinct types of band inversions allowed by symmetry. Using these model Hamiltonians, which emphasize the physical nature of the allowed couplings, we establish the generic electronic phase diagram noncentrosymmetric Dirac semimetals and show that it generically includes phases with coexistent Weyl point nodes or Weyl line nodes. In particular, for one specific type of band inversion in sixfold symmetric systems we show that Weyl line nodes are always present. Based on first-principles calculations, we predict that BiPd$_2$O$_4$ is a noncentrosymmetric Dirac semimetal under 20 Gpa pressure and hosts topological type-II Dirac points on the fourfold rotation axis. Furthermore, we propose that the hexagonal polar alloy LiZnSb$_{x}$Bi$_{1-x}$ realizes a Dirac semimetal with coexistent Weyl points. Interestingly, the emergence and location of the Weyl points is highly tunable and can be controlled by the alloy concentration $x$. More generally, our results not only establish band-inverted noncentrosymmetric systems as a broad and versatile class of topological semimetals, but also provide a framework for studying the quantum nonlinear Hall effect and nonlinear optical properties in the Dirac semimetals., Comment: 17 pages, 10 figures

Carbon allotropes have a large family of materials with varieties of crystal structures and properties and can realize different topological phases. Using first principles calculations, we predict a new two-dimensional (2D) carbon allotrope, namely penta-octa-graphene, which consists of pentagonal and octagonal carbon rings. We find that penta-octa-graphene can host both type-I and type-II Dirac line nodes (DLNs). The band inversion between conduction and valence bands forms the type-I DLNs and the two highest valence bands form the type-II DLNs. We find that the type-I DLNs are robust to the biaxial strain and the type-II DLNs can be driven to type-I when applying over 3 $\%$ biaxial stretching strain. A lattice model based on the $\pi$ orbitals of carbons is derived to understand the coexistence mechanism of type-I and type-II DLNs in penta-octa-graphene. Possible realizations and characterizations of this penta-octa-graphene in the experiment are also discussed. Our findings shed new light on the study of the coexistence of multiple topological states in the 2D carbon allotropes., Comment: 8 pages, 4 figures, accepted for publication in Carbon(https://doi.org/10.1016/j.carbon.2019.11.083)

Despite the growing interest in topological materials, the difficulty of experimentally synthesizing and integrating them with other materials has been one of the main barriers restricting access to their unique properties. Recent advances in synthesizing metastable phases of crystalline materials can help to overcome this barrier and offer new platforms to experimentally study and manipulate band topology. Because III-V semiconductors have a wide range of functional material applications (including optoelectronic devices, light-emitting diodes, and highly efficient solar cells), and because Bi-doped III-V materials can be synthesized by ion plantation and ion-cutoff methods, we revisit the effect of bismuth substitution in metastable III-V semiconductors. Through first-principles calculation methods, we show that in wurtzite structure III-V materials, Bi substitution can lead to band inversion phenomena and induce nontrivial topological properties. Specifically, we identify that GaBi and InBi are Dirac-Weyl semimetals, characterized by the coexistence of Dirac points and Weyl points, and $\text{GaAs}_{0.5} \text{Bi}_{0.5}$, $\text{GaSb}_{0.5} \text{Bi}_{0.5}$, $\text{InSb}_{0.5} \text{Bi}_{0.5}$ are triple-point semimetals, characterized by two sets of "near Dirac" triple points on the Fermi level. These experimentally-accessible bismuth-based topological semimetals can be integrated into the large family of functional III-V materials for experimental studies of heterostructures and future optoelectronic applications., Comment: 8 pages, 9 figures