Search

In the fields of computer vision and robotics, accurate pixel-level correspondences are essential for enabling advanced tasks such as structure-from-motion and simultaneous localization and mapping. Recent correspondence pruning methods usually focus on learning local consistency through k-nearest neighbors, which makes it difficult to capture robust context for each correspondence. We propose CorrAdaptor, a novel architecture that introduces a dual-branch structure capable of adaptively adjusting local contexts through both explicit and implicit local graph learning. Specifically, the explicit branch uses KNN-based graphs tailored for initial neighborhood identification, while the implicit branch leverages a learnable matrix to softly assign neighbors and adaptively expand the local context scope, significantly enhancing the model's robustness and adaptability to complex image variations. Moreover, we design a motion injection module to integrate motion consistency into the network to suppress the impact of outliers and refine local context learning, resulting in substantial performance improvements. The experimental results on extensive correspondence-based tasks indicate that our CorrAdaptor achieves state-of-the-art performance both qualitatively and quantitatively. The code and pre-trained models are available at https://github.com/TaoWangzj/CorrAdaptor., Comment: 8 pages, 4 figures, accepted by ECAI

In this report, we present our solutions to the EgoVis Challenges in CVPR 2024, including five tracks in the Ego4D challenge and three tracks in the EPIC-Kitchens challenge. Building upon the video-language two-tower model and leveraging our meticulously organized egocentric video data, we introduce a novel foundation model called EgoVideo. This model is specifically designed to cater to the unique characteristics of egocentric videos and provides strong support for our competition submissions. In the Ego4D challenges, we tackle various tasks including Natural Language Queries, Step Grounding, Moment Queries, Short-term Object Interaction Anticipation, and Long-term Action Anticipation. In addition, we also participate in the EPIC-Kitchens challenge, where we engage in the Action Recognition, Multiple Instance Retrieval, and Domain Adaptation for Action Recognition tracks. By adapting EgoVideo to these diverse tasks, we showcase its versatility and effectiveness in different egocentric video analysis scenarios, demonstrating the powerful representation ability of EgoVideo as an egocentric foundation model. Our codebase and pretrained models are publicly available at https://github.com/OpenGVLab/EgoVideo., Comment: Champion solutions in the EgoVis CVPR 2024 workshop

An oscillating bound state is a phenomenon where excitations mediated by the continuum modes oscillate persistently. Although it is generated by the superposition of two bound states in the continuum (BICs), such phenomenon is said to be unique to giant atoms. We present the phenomenon of an oscillating bound state with an alternative waveguide QED system, which is a one-dimensional semi-infinite waveguide coupled to a \textit{point-like} quantum emitter. This \textit{point-like} quantum emitter is $\Lambda$-type quantum system with one transition driven by a classical field., Comment: 7 pages 4 figures

Half-Heusler alloys such as the (Zr,Hf)NiSn intermetallic compounds are important thermoelectric materials for converting waste heat into electricity. Reduced electrical resistivity at the hot interface between the half-Heusler material and a metal contact is critical for device performance, however this has yet to be achieved in practice. Recent experimental work suggests that a coherent interface between half-Heusler and full-Heusler compounds can form due to diffusion of transition metal atoms into the vacant sublattice of the half-Heusler lattice. We study theoretically the structural and electronic properties of such an interface using a first-principles based approach that combines {\it ab initio} calculations with macroscopic modeling. We find that the prototypical interface HfNi$_2$Sn/HfNiSn provides very low contact resistivity and almost ohmic behavior over a wide range of temperatures and doping levels. Given the potential of these interfaces to remain stable over a wide range of temperatures, our study suggests that full-Heuslers might provide nearly ideal electrical contacts to half-Heuslers that can be harnessed for efficient thermoelectric generator devices., Comment: 8 pages, 8 figures

Half-Heusler(HH) alloys have shown promising thermoelectric properties in the medium and high temperature range. To harness these material properties for thermoelectric applications, it is important to realize electrical contacts with low electrical contact resistivity. However, little is known about the detailed structural and electronic properties of such contacts, and the expected values of contact resistivity. Here, we employ atomistic ab initio calculations to study electrical contacts in a subclass of HH alloys consisting of the compounds HfCoSb, HfZrCoSb, and HfZrNiSn. By using Ag as a prototypical metal, we show that the termination of the HH material critically determines the presence or absence of strong deformations at the interface. Our study includes contacts to doped materials, and the results indicate that the p-type materials generally form ohmic contacts while the n-type materials have a small Schottky barrier. We calculate the temperature dependence of the contact resistivity in the low to medium temperature range and provide quantitative values that set lower limits for these systems., Comment: 20 pages, 7 figures + Suppl. Mat., accepted for publication in Phys. Rev. Materials

V-telluride superlattice thin films have shown promising performance for on-chip cooling devices. Recent experimental studies have indicated that device performance is limited by the metal/semiconductor electrical contacts. One challenge in realizing a low resistivity contacts is the absence of fundamental knowledge of the physical and chemical properties of interfaces between metal and V-telluride materials. Here we present a combination of experimental and theoretical efforts to understand, design and harness low resistivity contacts to V-tellurides. Ab initio calculations are used to explore the effects of interfacial structure and chemical compositions on the electrical contacts, and an ab initio based macroscopic model is employed to predict the fundamental limit of contact resistivity as a function of both carrier concentration and temperature. Under the guidance of theoretical studies, we develop an experimental approach to fabricate low resistivity metal contacts to V-telluride thin film superlattices, achieving a 100-fold reduction compared to previous work. Interfacial characterization and analysis using both scanning transmission electron microscopy and energy-dispersive x-ray spectroscopy show the unusual interfacial morphology and the potential for further improvement in contact resistivity. Finally, we harness the improved contacts to realize an improved high-performance thermoelectric cooling module., Comment: 36 pages, 11 figures (includes SI) to appear in Advanced Electronic Materials