arXiv subjects

Kan Chen

Publications and source records attributed to Kan Chen.

At least 73 records · Page 4Linked to original sources

arXiv2021

Unbiased Teacher for Semi-Supervised Object Detection

Semi-supervised learning, i.e., training networks with both labeled and unlabeled data, has made significant progress recently. However, existing works have primarily focused on image classification tasks and neglected object detection which requires more annotation effort. In this work, we revisit the Semi-Supervised Object Detection (SS-OD) and identify the pseudo-labeling bias issue in SS-OD. To address this, we introduce Unbiased Teacher, a simple yet effective approach that jointly trains a student and a gradually progressing teacher in a mutually-beneficial manner. Together with a class-balance loss to downweight overly confident pseudo-labels, Unbiased Teacher consistently improved state-of-the-art methods by significant margins on COCO-standard, COCO-additional, and VOC datasets. Specifically, Unbiased Teacher achieves 6.8 absolute mAP improvements against state-of-the-art method when using 1% of labeled data on MS-COCO, achieves around 10 mAP improvements against the supervised baseline when using only 0.5, 1, 2% of labeled data on MS-COCO.

cs.CV↗

arXiv2021

Spectrum of the doubly charmed molecular pentaquarks in chiral effective field theory

We perform a systematic study on the interactions of the $Σ_c^{(*)}D^{(*)}$ systems within the framework of chiral effective field theory. We introduce the contact term, one-pion-exchange and two-pion-exchange contributions to describe the short-, long-, and intermediate-range interactions. The low energy constants of the $Σ_c^{(*)}D^{(*)}$ systems are estimated from the $N\bar{N}$ scattering data by introducing a quark level Lagrangian. With three solutions of LECs, all the $Σ_c^{(*)}D^{(*)}$ systems with isospin $I=1/2$ can form bound states, in which different inputs of LECs may lead to distinguishable mass spectra. In addition, we also investigate the interactions of the charmed-bottom $Σ_c^{(*)}\bar{B}^{(*)}$, $Σ_b^{(*)}D^{(*)}$, and $Σ_b^{(*)}\bar{B}^{(*)}$ systems. Among the obtained bound states, the bindings become deeper when the reduced masses of the corresponding systems are heavier.

hep-ph↗

arXiv2020

CPARR: Category-based Proposal Analysis for Referring Relationships

The task of referring relationships is to localize subject and object entities in an image satisfying a relationship query, which is given in the form of \texttt{ }. This requires simultaneous localization of the subject and object entities in a specified relationship. We introduce a simple yet effective proposal-based method for referring relationships. Different from the existing methods such as SSAS, our method can generate a high-resolution result while reducing its complexity and ambiguity. Our method is composed of two modules: a category-based proposal generation module to select the proposals related to the entities and a predicate analysis module to score the compatibility of pairs of selected proposals. We show state-of-the-art performance on the referring relationship task on two public datasets: Visual Relationship Detection and Visual Genome.

cs.CV↗

arXiv2020

FBNetV2: Differentiable Neural Architecture Search for Spatial and Channel Dimensions

Differentiable Neural Architecture Search (DNAS) has demonstrated great success in designing state-of-the-art, efficient neural networks. However, DARTS-based DNAS's search space is small when compared to other search methods', since all candidate network layers must be explicitly instantiated in memory. To address this bottleneck, we propose a memory and computationally efficient DNAS variant: DMaskingNAS. This algorithm expands the search space by up to $10^{14}\times$ over conventional DNAS, supporting searches over spatial and channel dimensions that are otherwise prohibitively expensive: input resolution and number of filters. We propose a masking mechanism for feature map reuse, so that memory and computational costs stay nearly constant as the search space expands. Furthermore, we employ effective shape propagation to maximize per-FLOP or per-parameter accuracy. The searched FBNetV2s yield state-of-the-art performance when compared with all previous architectures. With up to 421$\times$ less search cost, DMaskingNAS finds models with 0.9% higher accuracy, 15% fewer FLOPs than MobileNetV3-Small; and with similar accuracy but 20% fewer FLOPs than Efficient-B0. Furthermore, our FBNetV2 outperforms MobileNetV3 by 2.6% in accuracy, with equivalent model size. FBNetV2 models are open-sourced at https://github.com/facebookresearch/mobile-vision.

cs.CV↗

arXiv2020

Video Object Grounding using Semantic Roles in Language Description

We explore the task of Video Object Grounding (VOG), which grounds objects in videos referred to in natural language descriptions. Previous methods apply image grounding based algorithms to address VOG, fail to explore the object relation information and suffer from limited generalization. Here, we investigate the role of object relations in VOG and propose a novel framework VOGNet to encode multi-modal object relations via self-attention with relative position encoding. To evaluate VOGNet, we propose novel contrasting sampling methods to generate more challenging grounding input samples, and construct a new dataset called ActivityNet-SRL (ASRL) based on existing caption and grounding datasets. Experiments on ASRL validate the need of encoding object relations in VOG, and our VOGNet outperforms competitive baselines by a significant margin.

cs.CV↗

arXiv2020

How efficient is the streaming instability in viscous protoplanetary disks?

The streaming instability is a popular candidate for planetesimal formation by concentrating dust particles to trigger gravitational collapse. However, its robustness against physical conditions expected in protoplanetary disks is unclear. In particular, particle stirring by turbulence may impede the instability. To quantify this effect, we develop the linear theory of the streaming instability with external turbulence modelled by gas viscosity and particle diffusion. We find the streaming instability is sensitive to turbulence, with growth rates becoming negligible for alpha-viscosity parameters $α\gtrsim \mathrm{St} ^{1.5}$, where $\mathrm{St}$ is the particle Stokes number. We explore the effect of non-linear drag laws, which may be applicable to porous dust particles, and find growth rates are modestly reduced. We also find that gas compressibility increase growth rates by reducing the effect of diffusion. We then apply linear theory to global models of viscous protoplanetary disks. For minimum-mass Solar nebula disk models, we find the streaming instability only grows within disk lifetimes beyond $\sim 10$s of AU, even for cm-sized particles and weak turbulence ($α\sim 10^{-4}$). Our results suggest it is rather difficult to trigger the streaming instability in non-laminar protoplanetary disks, especially for small particles.

astro-ph.EP↗

arXiv2019

$\bar{K}Λ$ molecular explanation to the newly observed $Ξ(1620)^0$

The newly observed $Ξ(1620)^0$ by the Belle Collaboration inspires our interest in performing a systematic study on the interaction of an anti-strange meson $(\bar{K}^{(*)})$ with a strange or doubly strange ground octet baryon $\mathcal{B}$ ($Λ$, $Σ$, and $Ξ$), where the spin-orbit force and the recoil correction are considered in the adopted one-boson-exchange model. Our results indicate that $Ξ(1620)^0$ can be explained as a $\bar{K}Λ$ molecular state with $I(J^P)=1/2(1/2^-)$ and the intermediate force from $σ$ exchange plays an important role. Additionally, we also predict several other possible molecular candidates, i.e., the $\bar{K}Σ$ molecular state with $I(J^P)=1/2(1/2^-)$ and the triply strange $\bar{K}Ξ$ molecular state with $I(J^P)=0(1/2^-)$.

hep-ph↗

arXiv2019

Zero-Shot Grounding of Objects from Natural Language Queries

A phrase grounding system localizes a particular object in an image referred to by a natural language query. In previous work, the phrases were restricted to have nouns that were encountered in training, we extend the task to Zero-Shot Grounding(ZSG) which can include novel, "unseen" nouns. Current phrase grounding systems use an explicit object detection network in a 2-stage framework where one stage generates sparse proposals and the other stage evaluates them. In the ZSG setting, generating appropriate proposals itself becomes an obstacle as the proposal generator is trained on the entities common in the detection and grounding datasets. We propose a new single-stage model called ZSGNet which combines the detector network and the grounding system and predicts classification scores and regression parameters. Evaluation of ZSG system brings additional subtleties due to the influence of the relationship between the query and learned categories; we define four distinct conditions that incorporate different levels of difficulty. We also introduce new datasets, sub-sampled from Flickr30k Entities and Visual Genome, that enable evaluations for the four conditions. Our experiments show that ZSGNet achieves state-of-the-art performance on Flickr30k and ReferIt under the usual "seen" settings and performs significantly better than baseline in the zero-shot setting.

cs.CV↗

arXiv2019

Cascaded Parallel Filtering for Memory-Efficient Image-Based Localization

Image-based localization (IBL) aims to estimate the 6DOF camera pose for a given query image. The camera pose can be computed from 2D-3D matches between a query image and Structure-from-Motion (SfM) models. Despite recent advances in IBL, it remains difficult to simultaneously resolve the memory consumption and match ambiguity problems of large SfM models. In this work, we propose a cascaded parallel filtering method that leverages the feature, visibility and geometry information to filter wrong matches under binary feature representation. The core idea is that we divide the challenging filtering task into two parallel tasks before deriving an auxiliary camera pose for final filtering. One task focuses on preserving potentially correct matches, while another focuses on obtaining high quality matches to facilitate subsequent more powerful filtering. Moreover, our proposed method improves the localization accuracy by introducing a quality-aware spatial reconfiguration method and a principal focal length enhanced pose estimation method. Experimental results on real-world datasets demonstrate that our method achieves very competitive localization performances in a memory-efficient manner.

cs.CV↗

arXiv2019

Billion-scale semi-supervised learning for image classification

This paper presents a study of semi-supervised learning with large convolutional networks. We propose a pipeline, based on a teacher/student paradigm, that leverages a large collection of unlabelled images (up to 1 billion). Our main goal is to improve the performance for a given target architecture, like ResNet-50 or ResNext. We provide an extensive analysis of the success factors of our approach, which leads us to formulate some recommendations to produce high-accuracy models for image classification with semi-supervised learning. As a result, our approach brings important gains to standard architectures for image, video and fine-grained classification. For instance, by leveraging one billion unlabelled images, our learned vanilla ResNet-50 achieves 81.2% top-1 accuracy on the ImageNet benchmark.

cs.CV↗

arXiv2018

Estimating the production rates of D-wave charmed mesons via the semileptonic decays of bottom mesons

In this work, using the covariant light front approach with conventional vertex functions, we estimate the production rates of $D$-wave charmed/charmed-strange mesons via $B_{(s)}$ semileptonic decays. Due to these calculated considerable production rates, it is possible to experimentally search for $D$-wave charmed/charmed-strange mesons via the semileptonic decays, which may provide extra approach to explore $D$-wave charmed/charmed-strange mesons.

hep-ph↗

arXiv2018

MAC: Mining Activity Concepts for Language-based Temporal Localization

We address the problem of language-based temporal localization in untrimmed videos. Compared to temporal localization with fixed categories, this problem is more challenging as the language-based queries not only have no pre-defined activity list but also may contain complex descriptions. Previous methods address the problem by considering features from video sliding windows and language queries and learning a subspace to encode their correlation, which ignore rich semantic cues about activities in videos and queries. We propose to mine activity concepts from both video and language modalities by applying the actionness score enhanced Activity Concepts based Localizer (ACL). Specifically, the novel ACL encodes the semantic concepts from verb-obj pairs in language queries and leverages activity classifiers' prediction scores to encode visual concepts. Besides, ACL also has the capability to regress sliding windows as localization results. Experiments show that ACL significantly outperforms state-of-the-arts under the widely used metric, with more than 5% increase on both Charades-STA and TACoS datasets.

cs.CV↗

arXiv2018

Surveying exotic pentaquarks with the typical $QQqq\bar{q}$ configuration

As a hot issue, exploring exotic pentaquarks is full of challenges and opportunities for both theorist and experimentalist. In this work, we focus on a type of pentaquark with the $QQqq\bar{q}$ ($Q=b,c$; $q=u,d,s$) configuration, where their mass spectrum is estimated systematically. Especially, our result indicates that there may exist some stable or narrow exotic pentaquark states. Obviously, our study may provides valuable information for further experimental search for the $QQqq\bar{q}$ pentaquarks. With the running of LHCb and forthcoming Belle II, we have a reason to believe that these predictions present here can be tested.

hep-ph↗

arXiv2018

CTAP: Complementary Temporal Action Proposal Generation

Temporal action proposal generation is an important task, akin to object proposals, temporal action proposals are intended to capture "clips" or temporal intervals in videos that are likely to contain an action. Previous methods can be divided to two groups: sliding window ranking and actionness score grouping. Sliding windows uniformly cover all segments in videos, but the temporal boundaries are imprecise; grouping based method may have more precise boundaries but it may omit some proposals when the quality of actionness score is low. Based on the complementary characteristics of these two methods, we propose a novel Complementary Temporal Action Proposal (CTAP) generator. Specifically, we apply a Proposal-level Actionness Trustworthiness Estimator (PATE) on the sliding windows proposals to generate the probabilities indicating whether the actions can be correctly detected by actionness scores, the windows with high scores are collected. The collected sliding windows and actionness proposals are then processed by a temporal convolutional neural network for proposal ranking and boundary adjustment. CTAP outperforms state-of-the-art methods on average recall (AR) by a large margin on THUMOS-14 and ActivityNet 1.3 datasets. We further apply CTAP as a proposal generation method in an existing action detector, and show consistent significant improvements.

cs.CV↗

arXiv2018

Heavy-flavored tetraquark states with the $QQ\bar{Q}\bar{Q}$ configuration

In the framework of the color-magnetic interaction, we systematically investigate the mass spectrum of the tetraquark states composed of four heavy quarks with the $QQ\bar Q\bar Q$ configuration in this work. We also show their strong decay patterns. Stable or narrow states in the $bb\bar{b}\bar{c}$ and $bc\bar{b}\bar{c}$ systems are found to be possible. We hope the studies shall be helpful to the experimental search for heavy-full exotic tetraquark states.

hep-ph↗

arXiv2018

Motion-Appearance Co-Memory Networks for Video Question Answering

Video Question Answering (QA) is an important task in understanding video temporal structure. We observe that there are three unique attributes of video QA compared with image QA: (1) it deals with long sequences of images containing richer information not only in quantity but also in variety; (2) motion and appearance information are usually correlated with each other and able to provide useful attention cues to the other; (3) different questions require different number of frames to infer the answer. Based these observations, we propose a motion-appearance comemory network for video QA. Our networks are built on concepts from Dynamic Memory Network (DMN) and introduces new mechanisms for video QA. Specifically, there are three salient aspects: (1) a co-memory attention mechanism that utilizes cues from both motion and appearance to generate attention; (2) a temporal conv-deconv network to generate multi-level contextual facts; (3) a dynamic fact ensemble method to construct temporal representation dynamically for different questions. We evaluate our method on TGIF-QA dataset, and the results outperform state-of-the-art significantly on all four tasks of TGIF-QA.

cs.CV↗

arXiv2018

Knowledge Aided Consistency for Weakly Supervised Phrase Grounding

Given a natural language query, a phrase grounding system aims to localize mentioned objects in an image. In weakly supervised scenario, mapping between image regions (i.e., proposals) and language is not available in the training set. Previous methods address this deficiency by training a grounding system via learning to reconstruct language information contained in input queries from predicted proposals. However, the optimization is solely guided by the reconstruction loss from the language modality, and ignores rich visual information contained in proposals and useful cues from external knowledge. In this paper, we explore the consistency contained in both visual and language modalities, and leverage complementary external knowledge to facilitate weakly supervised grounding. We propose a novel Knowledge Aided Consistency Network (KAC Net) which is optimized by reconstructing input query and proposal's information. To leverage complementary knowledge contained in the visual features, we introduce a Knowledge Based Pooling (KBP) gate to focus on query-related proposals. Experiments show that KAC Net provides a significant improvement on two popular datasets.

cs.CV↗

arXiv2017

Regge-like relation and a universal description of heavy-light systems

Using the Regge-like formula $(M-m_Q)^2=πσL$ between hadron mass $M$ and angular momentum $L$ with a heavy quark mass $m_Q$ and a string tension $σ$, we analyze all the heavy-light systems, i.e., $D/D_s/B/B_s$ mesons and charmed and bottom baryons.Numerical plots are obtained for all the heavy-light mesons of experimental data whose slope becomes nearly equal to 1/2 of that for light hadrons. Assuming that charmed and bottom baryons consist of one heavy quark and one light cluster of two light quarks (diquark), we apply the formula to all the heavy-light baryons including recently discovered $Ω_c$'s and find that these baryons experimentally measured satisfy the above formula. We predict the average mass values of $B$, $B_s$, $Λ_b$, $Σ_c$, $Ξ_c$, and $Ω_c$ with $L=2$ as 6.01, 6.13, 6.15, 3.05, 3.07, and 3.34 GeV, respectively. Our results on baryons suggest that these baryons can be safely regarded as heavy quark-light cluster configuration. We also find a universal description for all the heavy-light mesons as well as baryons, i.e., one unique line is enough to describe both of charmed and bottom heavy-light systems. Our results suggest that instead of mass itself, gluon flux energy is essential to obtain a linear trajectory.

hep-ph↗