Search

We refine a recent result of Drury concerning the optimal ratio between the norm and numerical radius of a bounded linear operator $T$ with numerical range lying in a sector of a circular disk. In particular, characterization is given to the operators attaining the optimal ratio, and properties of such operators are explored., Comment: 15 pages

Learning from pseudo-labels that generated with VLMs~(Vision Language Models) has been shown as a promising solution to assist open vocabulary detection (OVD) in recent studies. However, due to the domain gap between VLM and vision-detection tasks, pseudo-labels produced by the VLMs are prone to be noisy, while the training design of the detector further amplifies the bias. In this work, we investigate the root cause of VLMs' biased prediction under the OVD context. Our observations lead to a simple yet effective paradigm, coded MarvelOVD, that generates significantly better training targets and optimizes the learning procedure in an online manner by marrying the capability of the detector with the vision-language model. Our key insight is that the detector itself can act as a strong auxiliary guidance to accommodate VLM's inability of understanding both the ``background'' and the context of a proposal within the image. Based on it, we greatly purify the noisy pseudo-labels via Online Mining and propose Adaptive Reweighting to effectively suppress the biased training boxes that are not well aligned with the target object. In addition, we also identify a neglected ``base-novel-conflict'' problem and introduce stratified label assignments to prevent it. Extensive experiments on COCO and LVIS datasets demonstrate that our method outperforms the other state-of-the-arts by significant margins. Codes are available at https://github.com/wkfdb/MarvelOVD, Comment: Codes are available at https://github.com/wkfdb/MarvelOVD

Semi-Supervised Object Detection (SSOD) has achieved resounding success by leveraging unlabeled data to improve detection performance. However, in Open Scene Semi-Supervised Object Detection (O-SSOD), unlabeled data may contains unknown objects not observed in the labeled data, which will increase uncertainty in the model's predictions for known objects. It is detrimental to the current methods that mainly rely on self-training, as more uncertainty leads to the lower localization and classification precision of pseudo labels. To this end, we propose Credible Teacher, an end-to-end framework. Credible Teacher adopts an interactive teaching mechanism using flexible labels to prevent uncertain pseudo labels from misleading the model and gradually reduces its uncertainty through the guidance of other credible pseudo labels. Empirical results have demonstrated our method effectively restrains the adverse effect caused by O-SSOD and significantly outperforms existing counterparts., Comment: Accpet by ICASSP 2024

Planetary Radio Interferometry and Doppler Experiment (PRIDE) is a multi-purpose experimental technique aimed at enhancing the science return of planetary missions. The technique exploits the science payload and spacecraft service systems without requiring a dedicated onboard instrumentation or imposing on the existing instrumentation any special for PRIDE requirements. PRIDE is based on the near-field phase-referencing Very Long Baseline Interferometry (VLBI) and evaluation of the Doppler shift of the radio signal transmitted by spacecraft by observing it with multiple Earth-based radio telescopes. The methodology of PRIDE has been developed initially at the Joint Institute for VLBI ERIC (JIVE) for tracking the ESA's Huygens Probe during its descent in the atmosphere of Titan in 2005. From that point on, the technique has been demonstrated for various planetary and other space science missions. The estimates of lateral position of the target spacecraft are done using the phase-referencing VLBI technique. Together with radial Doppler estimates, these observables can be used for a variety of applications, including improving the knowledge of the spacecraft state vector. The PRIDE measurements can be applied to a broad scope of research fields including studies of atmospheres through the use of radio occultations, the improvement of planetary and satellite ephemerides, as well as gravity field parameters and other geodetic properties of interest, and estimations of interplanetary plasma properties. This paper presents the implementation of PRIDE as a component of the ESA's Jupiter Icy Moons Explorer (JUICE) mission., Comment: Accepted for publication in Space Science Reviews, v. 219 (special issue on the ESA's JUICE mission)

The supernova remnant (SNR) 30 Dor B is associated with the \ion{H}{2} region ionized by the OB association LH99. The complex interstellar environment has made it difficult to study the physical structure of this SNR. We have used Hubble Space Telescope H$\alpha$ images to identify SNR shocks and deep Chandra X-ray observations to detect faint diffuse emission. We find that 30 Dor B hosts three zones with very different X-ray surface brightnesses and nebular kinematics that are characteristic of SNRs in different interstellar environments and/or evolutionary stages. The ASKAP 888 MHz map of 30 Dor B shows counterparts to all X-ray emission features except the faint halo. The ASKAP 888 MHz and 1420 MHz observations are used to produce a spectral index map, but its interpretation is complicated by the background thermal emission and the pulsar PSR J0537$-$6910's flat spectral index. The stellar population in the vicinity of 30 Dor B indicates a continuous star formation in the past 8--10 Myr. The observed very massive stars in LH99 cannot be coeval with the progenitor of 30 Dor B's pulsar. Adopting the pulsar's spin-down timescale, 5000 yr, as the age of the SNR, the X-ray shell would be expanding at $\sim$4000 km\,s$^{-1}$ and the post-shock temperature would be 1--2 orders of magnitude higher than that indicated by the X-ray spectra. Thus, the bright central region of 30 Dor B and the X-ray shell requires two separate SN events, and the faint diffuse X-ray halo perhaps other older SN events.

We analyze the DETR-based framework on semi-supervised object detection (SSOD) and observe that (1) the one-to-one assignment strategy generates incorrect matching when the pseudo ground-truth bounding box is inaccurate, leading to training inefficiency; (2) DETR-based detectors lack deterministic correspondence between the input query and its prediction output, which hinders the applicability of the consistency-based regularization widely used in current SSOD methods. We present Semi-DETR, the first transformer-based end-to-end semi-supervised object detector, to tackle these problems. Specifically, we propose a Stage-wise Hybrid Matching strategy that combines the one-to-many assignment and one-to-one assignment strategies to improve the training efficiency of the first stage and thus provide high-quality pseudo labels for the training of the second stage. Besides, we introduce a Crossview Query Consistency method to learn the semantic feature invariance of object queries from different views while avoiding the need to find deterministic query correspondence. Furthermore, we propose a Cost-based Pseudo Label Mining module to dynamically mine more pseudo boxes based on the matching cost of pseudo ground truth bounding boxes for consistency training. Extensive experiments on all SSOD settings of both COCO and Pascal VOC benchmark datasets show that our Semi-DETR method outperforms all state-of-the-art methods by clear margins. The PaddlePaddle version code1 is at https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/semi_det/semi_detr., Comment: CVPR2023

The EnVision and VERITAS missions to Venus will fly with X and Ka band telecommunications channels which can be used to conduct radio occultation studies of Venus' atmosphere. While link attenuation measurements during prior S and X band occultation experiments have been used to determine vertical profiles of H$_2$SO$_4$ vapor abundance, the addition of the Ka band channel introduces greater sensitivity to the abundances of H$_2$SO$_4$ aerosols and SO$_2$ gas, permitting retrieval of their vertical profiles from dual band measurements. Such measurements would be valuable in the assessment of chemical and dynamical processes governing short and long-term variability in Venus' atmosphere. This paper considers the sensitivity of the X/Ka band radio attenuation measurement to these atmospheric constituents, as well as uncertainties and regularization approaches for conducting retrievals of these atmospheric sulfur species from future occultation experiments. We introduce methods for seeding maximum likelihood estimation retrievals using shape models and simple atmospheric transport constraints. From simulated retrievals, we obtain mean errors of the order of 0.5 ppm, 20 ppm, and 10 mg/m$^3$ for H$_2$SO$_4$ vapor, SO$_2$, and H$_2$SO$_4$ aerosol abundances, respectively, for simultaneous retrieval., Comment: 29 pages, 11 figures, accepted to PSJ

The prediction of traffic flow is a challenging yet crucial problem in spatial-temporal analysis, which has recently gained increasing interest. In addition to spatial-temporal correlations, the functionality of urban areas also plays a crucial role in traffic flow prediction. However, the exploration of regional functional attributes mainly focuses on adding additional topological structures, ignoring the influence of functional attributes on regional traffic patterns. Different from the existing works, we propose a novel module named POI-MetaBlock, which utilizes the functionality of each region (represented by Point of Interest distribution) as metadata to further mine different traffic characteristics in areas with different functions. Specifically, the proposed POI-MetaBlock employs a self-attention architecture and incorporates POI and time information to generate dynamic attention parameters for each region, which enables the model to fit different traffic patterns of various areas at different times. Furthermore, our lightweight POI-MetaBlock can be easily integrated into conventional traffic flow prediction models. Extensive experiments demonstrate that our module significantly improves the performance of traffic flow prediction and outperforms state-of-the-art methods that use metadata., Comment: 15 pages, 8 figures.This work has been accepted by Information Sciences

Science archives are cornerstones of modern astronomical facilities. In this paper we describe the version 1.0 milestone of the Atacama Large Millimeter/submillimeter Array Science Archive. This version features a comprehensive query interface with rich metadata and visualisation of the spatial and spectral locations of the observations, a complete set of virtual observatory services for programmatic access, text-based similarity search, display and query for types of astronomical objects in SIMBAD and NED, browser-based remote visualisation, interactive previews with tentative line identification and extensive documentation including video and Jupyter Notebook tutorials. The development is regularly evaluated by means of user surveys and is entirely focused on providing the best possible user experience with the goal of helping to maximise the scientific productivity of the observatory., Comment: 6 pages, 7 figues

Land remote sensing analysis is a crucial research in earth science. In this work, we focus on a challenging task of land analysis, i.e., automatic extraction of traffic roads from remote sensing data, which has widespread applications in urban development and expansion estimation. Nevertheless, conventional methods either only utilized the limited information of aerial images, or simply fused multimodal information (e.g., vehicle trajectories), thus cannot well recognize unconstrained roads. To facilitate this problem, we introduce a novel neural network framework termed Cross-Modal Message Propagation Network (CMMPNet), which fully benefits the complementary different modal data (i.e., aerial images and crowdsourced trajectories). Specifically, CMMPNet is composed of two deep Auto-Encoders for modality-specific representation learning and a tailor-designed Dual Enhancement Module for cross-modal representation refinement. In particular, the complementary information of each modality is comprehensively extracted and dynamically propagated to enhance the representation of another modality. Extensive experiments on three real-world benchmarks demonstrate the effectiveness of our CMMPNet for robust road extraction benefiting from blending different modal data, either using image and trajectory data or image and Lidar data. From the experimental results, we observe that the proposed approach outperforms current state-of-the-art methods by large margins.Our source code is resealed on the project page http://lingboliu.com/multimodal_road_extraction.html., Comment: This work has been accepted by IEEE Transactions on Neural Networks and Learning Systems