编辑丨极市平台 CVPR2023已经放榜,今年有2360篇,接收率为25。78。在CVPR2023正式会议召开前,为了让大家更快地获取和学习到计算机视觉前沿技术,极市对CVPR2023最新论文进行追踪,包括分研究方向的论文、代码汇总以及论文技术直播分享。 CVPR2023论文分方向整理目前在极市社区持续更新中,已累计更新了381篇,项目地址:https:www。cvmart。netcommunitydetail7422 以下是最近更新的CVPR2023论文,包含检测、分割、人脸、视频处理、医学影像、神经网络结构、多模态、小样本学习等方向。 下载地址:https:www。cvmart。netcommunitydetail7454目录检测分割视频处理估计人脸目标跟踪图像视频检索视频理解医学影像GAN生成式对抗式图像生成图像合成神经网络结构设计数据处理模型训练泛化图像特征提取与匹配视觉表征学习模型评估多模态学习视觉预测数据集小样本学习零样本学习持续学习迁移学习domain自适应场景图视觉定位位姿估计视觉推理视觉问答对比学习强化学习机器人半监督学习弱监督学习无监督学习自监督学习其他检测2D目标检测(2DObjectDetection) 〔1〕ObjectAwareDistillationPyramidforOpenVocabularyObjectDetection paper:https:arxiv。orgabs2303。058923D目标检测(3Dobjectdetection) 〔1〕Bi3D:BidomainActiveLearningforCrossdomain3DObjectDetection paper:https:arxiv。orgabs2303。05886 〔2〕PiMAE:PointCloudandImageInteractiveMaskedAutoencodersfor3DObjectDetection paper:https:arxiv。orgabs2303。08129 code:https:github。comblvlabpimae 〔3〕MSF:MotionguidedSequentialFusionforEfficient3DObjectDetectionfromPointCloudSequences paper:https:arxiv。orgabs2303。08316 〔4〕CAPE:CameraViewPositionEmbeddingforMultiView3DObjectDetection paper:https:arxiv。orgabs2303。10209 code:https:github。comPaddlePaddlePaddle3D 〔5〕WeaklySupervisedMonocular3DObjectDetectionusingMultiViewProjectionandDirectionConsistency paper:https:arxiv。orgabs2303。08686) 〔6〕AeDet:AzimuthinvariantMultiview3DObjectDetection paper:https:arxiv。orgabs2211。12501 code:https:github。comfcjianAeDet异常检测(AnomalyDetection) 〔1〕DeSTSeg:SegmentationGuidedDenoisingStudentTeacherforAnomalyDetection paper:https:arxiv。orgabs2211。11317分割全景分割(PanopticSegmentation) 〔1〕UniDAformer:UnifiedDomainAdaptivePanopticSegmentationTransformerviaHierarchicalMaskCalibration paper:https:arxiv。orgabs2206。15083 语义分割(SemanticSegmentation) 〔1〕MSeg3D:Multimodal3DSemanticSegmentationforAutonomousDriving paper:https:arxiv。orgabs2303。08600 code:https:github。comjialeli1lidarseg3d 〔2〕SideAdapterNetworkforOpenVocabularySemanticSegmentation paper:https:arxiv。orgabs2302。12242 code:https:github。commendelxusan 〔3〕MultiviewInverseRenderingforLargescaleRealworldIndoorScenes paper:https:arxiv。orgabs2211。10206实例分割(InstanceSegmentation) 〔1〕FastInst:ASimpleQueryBasedModelforRealTimeInstanceSegmentation paper:https:arxiv。orgabs2303。08594 〔2〕SIM:SemanticawareInstanceMaskGenerationforBoxSupervisedInstanceSegmentation paper:https:arxiv。orgabs2303。08578 code:https:github。comlslrhsim 〔3〕DynaMask:DynamicMaskSelectionforInstanceSegmentation paper:https:arxiv。orgabs2303。07868 code:https:github。comlslrhdynamask视频目标分割(VideoObjectSegmentation) 〔1〕MobileVOS:RealTimeVideoObjectSegmentationContrastiveLearningmeetsKnowledgeDistillation paper:https:arxiv。orgabs2303。07815 〔2〕InstMove:InstanceMotionforObjectcentricVideoSegmentation paper:https:arxiv。orgabs2303。08132 code:https:github。comwjf5203vnext 〔3〕UnifiedMaskEmbeddingandCorrespondenceLearningforSelfSupervisedVideoSegmentation paper:https:arxiv。orgabs2303。10100视频处理(VideoProcessing) 〔1〕MobileVOS:RealTimeVideoObjectSegmentationContrastiveLearningmeetsKnowledgeDistillation paper:https:arxiv。orgabs2303。07815 〔2〕InstMove:InstanceMotionforObjectcentricVideoSegmentation paper:https:arxiv。orgabs2303。08132 code:https:github。comwjf5203vnext 〔3〕VideoDehazingviaaMultiRangeTemporalAlignmentNetworkwithPhysicalPrior paper:https:arxiv。orgabs2303。09757 code:https:github。comjiaqixuacmapnet 〔4〕BlindVideoDeflickeringbyNeuralFilteringwithaFlawedAtlas paper:https:arxiv。orgabs2303。08120 code:https:github。comchenyangleiallinonedeflicker视频生成视频合成(VideoGenerationVideoSynthesis) 〔1〕3DCinemagraphyfromaSingleImage paper:https:arxiv。orgabs2303。05724 〔2〕VideoFusion:DecomposedDiffusionModelsforHighQualityVideoGeneration paper:https:arxiv。orgabs2303。08320 code:https:github。commodelscopemodelscope视频超分(VideoSuperResolution) 〔1〕TowardsHighQualityandEfficientVideoSuperResolutionviaSpatialTemporalDataOverfitting paper:https:arxiv。orgabs2303。08331估计光流运动估计(OpticalFlowMotionEstimation) 〔1〕RethinkingOpticalFlowfromGeometricMatchingConsistentPerspective paper:https:arxiv。orgabs2303。08384 code:https:github。comdqiaolematchflow 深度估计(DepthEstimation) 〔1〕FullySelfSupervisedDepthEstimationfromDefocusClue paper:https:arxiv。orgabs2303。10752 code:https:github。comehzoahisdered人体解析人体姿态估计(HumanParsingHumanPoseEstimation) 〔1〕MutualInformationBasedTemporalDifferenceLearningforHumanPoseEstimationinVideo paper:https:arxiv。orgabs2303。08475 〔2〕MarkerlessCameratoRobotPoseEstimationviaSelfsupervisedSimtoRealTransfer paper:https:arxiv。orgabs2302。14338手势估计(GestureEstimation) 〔1〕CVTSLR:ContrastiveVisualTextualTransformationforSignLanguageRecognitionwithVariationalAlignment paper:https:arxiv。orgabs2303。05725 code:https:arxiv。orgabs2303。05725图像处理 〔1〕DeltaEdit:ExploringTextfreeTrainingforTextDrivenImageManipulation paper:https:arxiv。orgabs2303。06285 code:https:github。comyueming6568deltaedit图像复原图像增强图像重建(ImageRestorationImageReconstruction) 〔1〕ContrastiveSemisupervisedLearningforUnderwaterImageRestorationviaReliableBank paper:https:arxiv。orgabs2303。09101 code:https:github。comhuangshiruisemiuir 〔1〕ACR:AttentionCollaborationbasedRegressorforArbitraryTwoHandReconstruction paper:https:arxiv。orgabs2303。05938 code:https:github。comzhengdiyuarbitraryhands3dreconstruction 风格迁移(StyleTransfer) 〔1〕StyleRF:Zeroshot3DStyleTransferofNeuralRadianceFields paper:https:arxiv。orgabs2303。10598 〔2〕FixtheNoise:DisentanglingSourceFeatureforTransferLearningofStyleGAN paper:https:arxiv。orgabs2204。14079 code:https:github。comLeeDongYeunFixNoise人脸人脸识别检测(FacialRecognitionDetection) 〔1〕LocalRegionPerceptionandRelationshipLearningCombinedwithFeatureFusionforFacialActionUnitDetection paper:https:arxiv。orgabs2303。08545 〔2〕MultiModalFacialExpressionRecognitionwithTransformerBasedFusionNetworksandDynamicSampling paper:https:arxiv。orgabs2303。08419人脸生成合成重建编辑(FaceGenerationFaceSynthesisFaceReconstructionFaceEditing) 〔1〕RobustModelbasedFaceReconstructionthroughWeaklySupervisedOutlierSegmentation paper:https:arxiv。orgabs2106。09614 code:https:github。comunibasgravisOcclusionRobustMoFA目标跟踪(ObjectTracking) 〔1〕MotionTrack:LearningRobustShorttermandLongtermMotionsforMultiObjectTracking paper:https:arxiv。orgabs2303。10404 〔2〕VisualPromptMultiModalTracking paper:https:arxiv。orgabs2303。10826 code:https:github。comjiawenzhuvipt图像视频检索视频理解(ImageVideoRetrievalVideoUnderstanding) 〔1〕DataFreeSketchBasedImageRetrieval paper:https:arxiv。orgabs2303。07775 〔2〕DAA:ADeltaAgeAdaINoperationforageestimationviabinarycodetransformer paper:https:arxiv。orgabs2303。07929 〔3〕DualpathAdaptationfromImagetoVideoTransformers paper:https:arxiv。orgabs2303。09857 code:https:github。comparkjungindualpath图像视频字幕(ImageVideoCaption) 〔1〕DualStreamTransformerforGenericEventBoundaryCaptioning paper:https:arxiv。orgabs2207。03038 code:https:github。comgx77dualstreamtransformerforgenericeventboundarycaptioning行为识别动作识别检测分割定位(ActionActivityRecognition) 〔1〕VideoTestTimeAdaptationforActionRecognition paper:https:arxiv。orgabs2211。15393行人重识别检测(ReIdentificationDetection) 〔1〕TranSG:TransformerBasedSkeletonGraphPrototypeContrastiveLearningwithStructureTrajectoryPromptedReconstructionforPersonReIdentification paper:https:arxiv。orgabs2303。06819 code:https:github。comkalihactransg医学影像(MedicalImaging) 〔1〕NeuronStructureModelingforGeneralizableRemotePhysiologicalMeasurement paper:https:arxiv。orgabs2303。05955 code:https:github。comlupaopaonest 〔2〕UnsupervisedContourTrackingofLiveCellsbyMechanicalandCycleConsistencyLosses paper:https:arxiv。orgabs2303。08364 code:https:github。comjunbongjangcontourtracking 〔3〕TaskspecificFinetuningviaVariationalInformationBottleneckforWeaklysupervisedPathologyWholeSlideImageClassification paper:https:arxiv。orgabs2303。08446GAN生成式对抗式(GANGenerativeAdversarial) 〔2〕GraphTransformerGANsforGraphConstrainedHouseGeneration paper:https:arxiv。orgabs2303。08225 〔1〕CrossGANAuditing:UnsupervisedIdentificationofAttributeLevelSimilaritiesandDifferencesbetweenPretrainedGenerativeModels paper:https:arxiv。orgabs2303。10774图像生成图像合成(ImageGenerationImageSynthesis) 〔1〕3DQD:GeneralizedDeep3DShapePriorviaPartDiscretizedDiffusionProcess paper:https:arxiv。orgabs2303。10406 code:https:github。comcolorfulliyu3dqd 〔2〕ADynamicMultiScaleVoxelFlowNetworkforVideoPrediction paper:https:arxiv。orgabs2303。09875 code:https:github。commegviiresearchCVPR2023DMVFN 〔3〕RegularizedVectorQuantizationforTokenizedImageSynthesis paper:https:arxiv。orgabs2303。06424三维视觉点云(PointCloud) 〔1〕ControllableMeshGenerationThroughSparseLatentPointDiffusionModels paper:https:arxiv。orgabs2303。07938 〔2〕ParameterisNotAllYouNeed:StartingfromNonParametricNetworksfor3DPointCloudAnalysis paper:https:arxiv。orgabs2303。08134 code:https:github。comzrrskywalkerpointnn 〔3〕RotationInvariantTransformerforPointCloudMatching paper:https:arxiv。orgabs2303。08231 〔4〕DeepGraphbasedSpatialConsistencyforRobustNonrigidPointCloudRegistration paper:https:arxiv。orgabs2303。09950 code:https:github。comqinzheng93graphscnet三维重建(3DReconstruction) 〔1〕MaskedWaveletRepresentationforCompactNeuralRadianceFields paper:https:arxiv。orgabs2212。09069 〔2〕DecouplingHumanandCameraMotionfromVideosintheWild paper:https:arxiv。orgabs2302。12827 〔3〕StructuralMultiplaneImage:BridgingNeuralViewSynthesisand3DReconstruction paper:https:arxiv。orgabs2303。05937 〔4〕NEF:NeuralEdgeFieldsfor3DParametricCurveReconstructionfromMultiviewImages paper:https:arxiv。orgabs2303。07653 〔5〕PartNeRF:GeneratingPartAwareEditable3DShapeswithout3DSupervision paper:https:arxiv。orgabs2303。09554 〔6〕SDFusion:Multimodal3DShapeCompletion,Reconstruction,andGeneration paper:https:arxiv。orgabs2212。04493 code:https:github。comyccyenchichengSDFusion场景重建视图合成新视角合成(NovelViewSynthesis) 〔1〕RobustDynamicRadianceFields paper:https:arxiv。orgabs2301。02239 〔2〕I2SDF:IntrinsicIndoorSceneReconstructionandEditingviaRaytracinginNeuralSDFs paper:https:arxiv。orgabs2303。07634 〔3〕MobileNeRF:ExploitingthePolygonRasterizationPipelineforEfficientNeuralFieldRenderingonMobileArchitectures paper:https:arxiv。orgabs2208。00277 code:https:github。comgoogleresearchjax3d神经网络结构设计(NeuralNetworkStructureDesign) 〔1〕LargeKernel3D:ScalingupKernelsin3DSparseCNNs paper:https:arxiv。orgabs2206。10555 code:https:github。comdvlabresearchlargekernel3dCNN 〔1〕RandomizedAdversarialTrainingviaTaylorExpansion paper:https:arxiv。orgabs2303。10653 code:https:github。comalexkaelrandomizedadversarialtraining 〔2〕AliasFreeConvnets:FractionalShiftInvarianceviaPolynomialActivations paper:https:arxiv。orgabs2303。08085 code:https:github。comhmichaelialiasfreeconvnetsTransformer 〔1〕BiFormer:VisionTransformerwithBiLevelRoutingAttention paper:https:arxiv。orgabs2303。08810 code:https:github。comrayleizhubiformer 〔2〕MakingVisionTransformersEfficientfromATokenSparsificationView paper:https:arxiv。orgabs2303。08685图神经网络(GNN) 〔1〕TurningStrengthsintoWeaknesses:ACertifiedRobustnessInspiredAttackFrameworkagainstGraphNeuralNetworks paper:https:arxiv。orgabs2303。06199数据处理 〔1〕TINC:TreestructuredImplicitNeuralCompression paper:https:arxiv。orgabs2211。06689 code:https:github。comrichealyoungtinc图像聚类(ImageClustering) 〔1〕OntheEffectsofSelfsupervisionandContrastiveAlignmentinDeepMultiviewClustering paper:https:arxiv。orgabs2303。09877 code:https:github。comdanieltrostendeepmvc模型训练泛化(ModelTrainingGeneralization) 〔1〕HumanBench:TowardsGeneralHumancentricPerceptionwithProjectorAssistedPretraining paper:https:arxiv。orgabs2303。05675 〔2〕UniversalInstancePerceptionasObjectDiscoveryandRetrieval paper:https:arxiv。orgabs2303。06674 code:https:github。comMasterBinIIAUUNINEXT 〔3〕SharpnessAwareGradientMatchingforDomainGeneralization paper:https:arxiv。orgabs2303。10353 code:https:github。comwangpengfeisagm图像特征提取与匹配(Imagefeatureextractionandmatching) 〔2〕IterativeGeometryEncodingVolumeforStereoMatching paper:https:arxiv。orgabs2303。06615 code:https:github。comgangweixigev 〔1〕ReferringImageMatting paper:https:arxiv。orgabs2206。05149 code:https:github。comjizhizilirim视觉表征学习(VisualRepresentationLearning) 〔1〕MARLIN:MaskedAutoencoderforfacialvideoRepresentationLearnINg paper:https:arxiv。orgabs2211。06627 code:https:github。comControlNetMARLIN模型评估(ModelEvaluation) 〔1〕TrojDiff:TrojanAttacksonDiffusionModelswithDiverseTargets paper:https:arxiv。orgabs2303。05762 code:https:github。comchenweixin107trojdiff多模态学习(MultiModalLearning) 〔1〕MutilmodalFeatureExtractionandAttentionbasedFusionforEmotionEstimationinVideos paper:https:arxiv。orgabs2303。10421 code:https:github。comxkwangcnabaw5thrtiai 〔2〕EmotionalReactionIntensityEstimationBasedonMultimodalData paper:https:arxiv。orgabs2303。09167 〔3〕MultimodalFeatureExtractionandFusionforEmotionalReactionIntensityEstimationandExpressionClassificationinVideoswithTransformers paper:https:arxiv。orgabs2303。09164 〔4〕UnderstandingandConstructingLatentModalityStructuresinMultimodalRepresentationLearning paper:https:arxiv。orgabs2303。05952视听学习(AudiovisualLearning) 〔1〕WatchorListen:RobustAudioVisualSpeechRecognitionwithVisualCorruptionModelingandReliabilityScoring paper:https:arxiv。orgabs2303。08536 code:https:github。comjoannahongavrelscore 〔2〕CASPNet:RethinkingVideoSaliencyPredictionfromanAudioVisualConsistencyPerceptualPerspective paper:https:arxiv。orgabs2303。06357 code:https:arxiv。orgabs2303。06357视觉语言(Visionlanguage) 〔1〕Lana:ALanguageCapableNavigatorforInstructionFollowingandGeneration paper:https:arxiv。orgabs2303。08409 code:https:github。comwxh1996lanavln视觉预测(VisionbasedPrediction) 〔1〕TBPFormer:LearningTemporalBirdsEyeViewPyramidforJointPerceptionandPredictioninVisionCentricAutonomousDriving paper:https:arxiv。orgabs2303。09998数据集(Dataset) 〔1〕AWhacAMoleDilemma:ShortcutsComeinMultiplesWhereMitigatingOneAmplifiesOthers paper:https:arxiv。orgabs2212。04825 code:https:github。comfacebookresearchWhacAMole 〔2〕MVImgNet:ALargescaleDatasetofMultiviewImages paper:https:arxiv。orgabs2303。06042 〔3〕SLOPER4D:ASceneAwareDatasetforGlobal4DHumanPoseEstimationinUrbanEnvironments paper:https:arxiv。orgabs2303。09095 code:https:github。comclimbingdailySLOPER4D 〔4〕AWhacAMoleDilemma:ShortcutsComeinMultiplesWhereMitigatingOneAmplifiesOthers paper:https:arxiv。orgabs2212。04825 code:https:github。comfacebookresearchWhacAMole 〔5〕MVImgNet:ALargescaleDatasetofMultiviewImages paper:https:arxiv。orgabs2303。06042小样本学习零样本学习(FewshotLearningZeroshotLearning) 〔1〕DiGeo:DiscriminativeGeometryAwareLearningforGeneralizedFewShotObjectDetection paper:https:arxiv。orgabs2303。09674 code:https:github。comphoenixvdigeo 〔2〕HubsandHyperspheres:ReducingHubnessandImprovingTransductiveFewshotLearningwithHypersphericalEmbeddings paper:https:arxiv。orgabs2303。09352 code:https:github。comuitmlnohub 〔3〕BidirectionalDistributionAlignmentforTransductiveZeroShotLearning paper:https:arxiv。orgabs2303。08698 code:https:github。comzhicaiwwwbivaegan持续学习(ContinualLearningLifelongLearning) 〔1〕AchievingaBetterStabilityPlasticityTradeoffviaAuxiliaryNetworksinContinualLearning paper:https:arxiv。orgabs2303。09483 code:https:github。comkimsanghwanancl迁移学习domain自适应(TransferLearningDomainAdaptation) 〔1〕TrainableProjectedGradientMethodforRobustFinetuning paper:https:arxiv。orgabs2303。10720 〔2〕DADETR:DomainAdaptiveDetectionTransformerwithInformationFusion paper:https:arxiv。orgabs2103。17084 〔3〕InstanceRelationGraphGuidedSourceFreeDomainAdaptiveObjectDetection paper:https:arxiv。orgabs2203。15793 code:https:github。comvibashanirgsfda 〔4〕InstanceRelationGraphGuidedSourceFreeDomainAdaptiveObjectDetection paper:https:arxiv。orgabs2203。15793 code:https:github。comvibashanirgsfda场景图场景图理解(SceneGraphUnderstanding) 〔1〕PLA:LanguageDrivenOpenVocabulary3DSceneUnderstanding paper:https:arxiv。orgabs2211。16312 code:https:github。comcvmilabpla视觉定位位姿估计(VisualLocalizationPoseEstimation) 〔1〕PSVT:EndtoEndMultiperson3DPoseandShapeEstimationwithProgressiveVideoTransformers paper:https:arxiv。orgabs2303。09187 〔2〕StructVPR:DistillStructuralKnowledgewithWeightingSamplesforVisualPlaceRecognition paper:https:arxiv。orgabs2212。00937视觉推理视觉问答(VisualReasoningVQA) 〔1〕DivideandConquer:AnsweringQuestionswithObjectFactorizationandCompositionalReasoning paper:https:arxiv。orgabs2303。10482 code:https:github。comszzexpoipoem 〔2〕GenerativeBiasforRobustVisualQuestionAnswering paper:https:arxiv。orgabs2208。00690对比学习(ContrastiveLearning) 〔1〕DynamicGraphEnhancedContrastiveLearningforChestXrayReportGeneration paper:https:arxiv。orgabs2303。10323 code:https:github。commlii0117dcl强化学习(ReinforcementLearning) 〔1〕EqMotion:EquivariantMultiagentMotionPredictionwithInvariantInteractionReasoning paper:https:arxiv。orgabs2303。10876 code:https:github。commediabrainsjtueqmotion机器人(Robotic) 〔1〕EfficientMapSparsificationBasedon2Dand3DDiscretizedGrids paper:https:arxiv。orgabs2303。10882半监督学习弱监督学习无监督学习自监督学习(SelfsupervisedLearningSemisupervisedLearning) 〔1〕ExtractingClassActivationMapsfromNonDiscriminativeFeaturesaswell paper:https:arxiv。orgabs2303。10334 code:https:github。comzhaozhengchenlpcam 〔2〕TeSLA:TestTimeSelfLearningWithAutomaticAdversarialAugmentation paper:https:arxiv。orgabs2303。09870 code:https:github。comdevavrattomartesla 〔3〕LOCATE:LocalizeandTransferObjectPartsforWeaklySupervisedAffordanceGrounding paper:https:arxiv。orgabs2303。09665 〔4〕MixTeacher:MiningPromisingLabelswithMixedScaleTeacherforSemiSupervisedObjectDetection paper:https:arxiv。orgabs2303。09061 code:https:github。comlliuzmixteacher 〔5〕SemisupervisedHandAppearanceRecoveryviaStructureDisentanglementandDualAdversarialDiscrimination paper:https:arxiv。orgabs2303。06380 〔6〕NonContrastiveUnsupervisedLearningofPhysiologicalSignalsfromVideo paper:https:arxiv。orgabs2303。07944其他 〔1〕FacialAffectiveAnalysisbasedonMAEandMultimodalInformationfor5thABAWCompetition paper:https:arxiv。orgabs2303。10849 〔2〕PartialNetworkCloning paper:https:arxiv。orgabs2303。10597 code:https:github。comjngwenyepncloning 〔3〕UncertaintyAwareOptimalTransportforSemanticallyCoherentOutofDistributionDetection paper:https:arxiv。orgabs2303。10449 code:https:github。comlufan31etood 〔4〕AdversarialCounterfactualVisualExplanations paper:https:arxiv。orgabs2303。09962 code:https:github。comguillaumejs2403ace 〔5〕ANewBenchmark:OntheUtilityofSyntheticDatawithBlenderforBareSupervisedLearningandDownstreamDomainAdaptation paper:https:arxiv。orgabs2303。09165 code:https:github。comhuitangtangontheutilityofsyntheticdata 〔6〕TamingDiffusionModelsforAudioDrivenCoSpeechGestureGeneration paper:https:arxiv。orgabs2303。09119 code:https:github。comadvocate99diffgesture 〔7〕SkinnedMotionRetargetingwithResidualPerceptionofMotionSemanticsGeometry paper:https:arxiv。orgabs2303。08658 code:https:github。comkebiir2et 〔8〕TowardsCompositionalAdversarialRobustness:GeneralizingAdversarialTrainingtoCompositeSemanticPerturbations paper:https:arxiv。orgabs2202。04235 code:https:github。comtwweebcompositeadv 〔9〕BackdoorDefenseviaDeconfoundedRepresentationLearning paper:https:arxiv。orgabs2303。06818 code:https:github。comzaixizhangcbd 〔10〕LabelInformationBottleneckforLabelEnhancement paper:https:arxiv。orgabs2303。06836 〔11〕LayoutDM:DiscreteDiffusionModelforControllableLayoutGeneration paper:https:arxiv。orgabs2303。08137 code:https:github。comCyberAgentAILablayoutdm 〔12〕DiversityAwareMetaVisualPrompting paper:https:arxiv。orgabs2303。08138 code:https:github。comshikiwdamvp