DiTing:A large-scale Chinese seismic benchmark dataset for artificial intelligence in seismology

In recent years,artificial intelligence technology has exhibited great potential in seismic signal recognition,setting off a new wave of research.Vast amounts of high-quality labeled data are required to develop and apply artificial intelligence in seismology research.In this study,based on the 2013–2020 seismic cataloging reports of the China Earthquake Networks Center,we constructed an artificial intelligence seismological training dataset(“DiTing”)with the largest known total time length.Data were recorded using broadband and short-period seismometers.The obtained dataset included 2,734,748 threecomponent waveform traces from 787,010 regional seismic events,the corresponding P-and S-phase arrival time labels,and 641,025 P-wave first-motion polarity labels.All waveforms were sampled at 50 Hz and cut to a time length of 180 s starting from a random number of seconds before the occurrence of an earthquake.Each three-component waveform contained a considerable amount of descriptive information,such as the epicentral distance,back azimuth,and signal-to-noise ratios.The magnitudes of seismic events,epicentral distance,signal-to-noise ratio of P-wave data,and signal-to-noise ratio of S-wave data ranged from 0 to 7.7,0 to 330 km,–0.05 to 5.31 dB,and–0.05 to 4.73 dB,respectively.The dataset compiled in this study can serve as a high-quality benchmark for machine learning model development and data-driven seismological research on earthquake detection,seismic phase picking,first-motion polarity determination,earthquake magnitude prediction,early warning systems,and strong ground-motion prediction.Such research will further promote the development and application of artificial intelligence in seismology.

QpefBD:A Benchmark Dataset Applied to Machine Learning for Minute-Scale Quantitative Precipitation Estimation and Forecasting

Nowcasts of strong convective precipitation and radar-based quantitative precipitation estimations have always been hot yet challenging issues in meteorological sciences.Data-driven machine learning,especially deep learning,provides a new technical approach for the quantitative estimation and forecasting of precipitation.A high-quality,large-sample,and labeled training dataset is critical for the successful application of machine-learning technology to a specific field.The present study develops a benchmark dataset that can be applied to machine learning for minutescale quantitative precipitation estimation and forecasting(QpefBD),containing 231,978 samples of 3185 heavy precipitation events that occurred in 6 provinces of central and eastern China from April to October 2016-2018.Each individual sample consists of 8 products of weather radars at 6-min intervals within the time window of the corresponding event and products of 27 physical quantities at hourly intervals that describe the atmospheric dynamic and thermodynamic conditions.Two data labels,i.e.,ground precipitation intensity and areal coverage of heavy precipitation at 6-min intervals,are also included.The present study describes the basic components of the dataset and data processing and provides metrics for the evaluation of model performance on precipitation estimation and forecasting.Based on these evaluation metrics,some simple and commonly used methods are applied to evaluate precipitation estimates and forecasts.The results can serve as the benchmark reference for the performance evaluation of machine learning models using this dataset.This paper also gives some suggestions and scenarios of the QpefBD application.We believe that the application of this benchmark dataset will promote interdisciplinary collaboration between meteorological sciences and artificial intelligence sciences,providing a new way for the identification and forecast of heavy precipitation.

卷烟包装外观缺陷数据集构建及深度学习检测技术研究

目的为了提升烟包缺陷检测的准确率,构建卷烟包装外观缺陷识别基准数据集,并开展主流深度学习模型在卷烟包装外观缺陷智能检测中的应用研究。方法首先,从生产运行中的ZB45型细支烟硬盒包装机组采集缺陷图像,经过人工审核与筛选后获取典型的缺陷数据。然后,根据缺陷的特征与成因,将缺陷数据划分为23个类别,并逐一进行目标检测框标注。最终,形成了包含13000余张缺陷图像的卷烟包装外观缺陷识别基准数据集,并针对烟包缺陷识别、缺陷分类、目标检测、模型迁移4项任务开展实验。结果结果表明,数据集能够满足高准确率深度学习模型的训练需求;通过模型迁移,能够利用该数据集大幅提高不同牌号卷烟的缺陷检测效果;DenseNet模型在烟包缺陷识别与缺陷分类任务上表现较好,准确率分别达到93.70%和95.43%,YOLOv5模型在缺陷目标检测任务上mAP@0.5值达到了96.61%。结论该数据集能够作为烟包缺陷检测领域的基准数据集,研究成果将进一步支撑卷烟包装领域的数据应用与数字化转型。

高光谱高空间分辨率遥感观测、处理与应用

高光谱遥感技术是遥感领域的研究热点之一。然而,由于成像口径与能量等限制因素,难以同时获得高光谱和高空间分辨率的图像,这极大限制了高光谱遥感在精细尺度任务中的应用。近年来,随着高光谱成像技术及无人机为代表的新型观测平台的发展,高光谱高空间(双高,同时具备纳米级光谱分辨率与亚米级空间分辨率)遥感技术发展迅猛,推动了高光谱遥感技术的应用,但同时也带来了更多问题。极高的空间与光谱分辨率使得数据更加海量高维,加剧了高光谱数据的空间异质性和光谱变异性,为影像智能信息处理带来更大的挑战。为此,本文将从双高遥感影像基准数据集、双高遥感影像智能信息处理、双高遥感影像典型应用3个方面论述双高遥感应用与发展现状。

Data Augmentation Using Contour Image for Convolutional Neural Network

With the development of artificial intelligence-related technologies such as deep learning,various organizations,including the government,are making various efforts to generate and manage big data for use in artificial intelligence.However,it is difficult to acquire big data due to various social problems and restrictions such as personal information leakage.There are many problems in introducing technology in fields that do not have enough training data necessary to apply deep learning technology.Therefore,this study proposes a mixed contour data augmentation technique,which is a data augmentation technique using contour images,to solve a problem caused by a lack of data.ResNet,a famous convolutional neural network(CNN)architecture,and CIFAR-10,a benchmark data set,are used for experimental performance evaluation to prove the superiority of the proposed method.And to prove that high performance improvement can be achieved even with a small training dataset,the ratio of the training dataset was divided into 70%,50%,and 30%for comparative analysis.As a result of applying the mixed contour data augmentation technique,it was possible to achieve a classification accuracy improvement of up to 4.64%and high accuracy even with a small amount of data set.In addition,it is expected that the mixed contour data augmentation technique can be applied in various fields by proving the excellence of the proposed data augmentation technique using benchmark datasets.

A Universal Activation Function for Deep Learning

Recently,deep learning has achieved remarkable results in fields that require human cognitive ability,learning ability,and reasoning ability.Activation functions are very important because they provide the ability of artificial neural networks to learn complex patterns through nonlinearity.Various activation functions are being studied to solve problems such as vanishing gradients and dying nodes that may occur in the deep learning process.However,it takes a lot of time and effort for researchers to use the existing activation function in their research.Therefore,in this paper,we propose a universal activation function(UA)so that researchers can easily create and apply various activation functions and improve the performance of neural networks.UA can generate new types of activation functions as well as functions like traditional activation functions by properly adjusting three hyperparameters.The famous Convolutional Neural Network(CNN)and benchmark datasetwere used to evaluate the experimental performance of the UA proposed in this study.We compared the performance of the artificial neural network to which the traditional activation function is applied and the artificial neural network to which theUA is applied.In addition,we evaluated the performance of the new activation function generated by adjusting the hyperparameters of theUA.The experimental performance evaluation results showed that the classification performance of CNNs improved by up to 5%through the UA,although most of them showed similar performance to the traditional activation function.

自然语言处理评测数据集质量评估研究

评测数据集是评测任务的载体,评测数据集的质量对评测任务的开展和评测指标的应用有着根本性的影响,因此对评测数据集的质量进行评估有着必要性和迫切性。该文在调研公开使用的自然语言处理主流数据集基础上,分析和总结了数据集中存在的8类问题,并在参考人类考试及试卷质量评估的基础上,从信度、效度和难度出发,提出了数据集评估的相关指标和将计算性与操作性相结合的评估方法,旨在为自然语言处理评测数据集构造、选择和使用提供参考依据。

蓝印花布纹样标准数据集的构建

针对在以数字化的方式传承和保护蓝印花布的技术中由于缺乏具有原始手工特征的蓝色印花图案数据集,极大地限制了深度学习技术在蓝印花布纹样识别领域的应用问题,构建了一个大规模蓝印花布纹样数据集。该数据集包含50 216张蓝印花布纹样,从动物、植物、中国神话传说及复合四大题材的角度将数据分为85个子类。该数据集的创建不仅涉及到蓝印花布的数字化建设,如蓝印花布纹样检索及相关纹样命名,还能使相关研究人员设计和验证机器学习算法。在所构建的数据集基础上,提供了4个典型深度学习网络的实验结果,以此作为其性能基准。

面向高光谱影像场景分类的轻量化深度全局-局部知识蒸馏网络

针对目标场景复杂的空间布局和高光谱影像固有的空-谱信息冗余等挑战,提出了端到端的轻量化深度全局-局部知识蒸馏(Lightweight Deep Global-Local Knowledge Distillation,LDGLKD)网络。为探索空-谱特征的全局序列属性,教师模型视觉Transformer(Vision Transformer,ViT)被用来指导轻量化学生模型进行高光谱影像场景分类。LDGLKD选择预训练的VGG16作为学生模型来提取局部细节信息,将ViT和VGG16通过知识蒸馏协同训练后,教师模型将所学习到的远程上下文关系向小规模学生模型进行传递。LDGLKD可通过知识蒸馏结合上述两种模型的优点,在欧比特高光谱影像场景分类数据集OHID-SC及公开的高光谱遥感图像数据集HSRS-SC上的最佳分类精度分别达到91.62%和97.96%。实验结果表明:LDGLKD网络具有良好的分类性能。根据欧比特珠海一号卫星提供的遥感数据构建的OHID-SC可以反映详细的地表覆盖情况,并为高光谱场景分类任务提供数据支撑。

Infrared-PV:面向监控应用的红外目标检测数据集

红外摄像机虽然能够全天候24 h工作,但是相比于可见光摄像机,其获得的红外图像分辨率和信杂比低,目标纹理信息缺乏,因此足够的标记图像和进行模型优化设计对于提高基于深度学习的红外目标检测性能具有重要意义。为解决面向监控应用场景的红外目标检测数据集缺乏的问题,首先采用红外摄像机采集了不同极性的红外图像,基于自研图像标注软件实现了VOC格式的图像标注任务,构建了一个包含行人和车辆两类目标的红外图像数据集(Infrared-PV),并对数据集中的目标特性进行了统计分析。然后采用主流的基于深度学习的目标检测模型进行了模型训练与测试,定性和定量分析了YOLO系列和Faster R-CNN系列等模型对于该数据集的目标检测性能。构建的红外目标数据集共包含图像2138张,场景中目标包含白热、黑热和热力图3种模式。当采用各模型进行目标检测性能测试时,Cascade R-CNN模型性能最优,mAP0.5值达到了82.3%,YOLO v5系列模型能够兼顾实时性和检测精度的平衡,推理速度达到175.4帧/s的同时mAP0.5值仅降低2.7%。构建的红外目标检测数据集能够为基于深度学习的红外图像目标检测模型优化研究提供一定的数据支撑,同时也可以用于目标的红外特性分析。

视觉场景描述及其效果评价

作为计算机视觉?多媒体?人工智能和自然语言处理等领域的交叉性研究课题,视觉场景描述的研究内容是自动生成一个或多个语句用于描述图像或视频中呈现的视觉场景信息.视觉场景中内容的丰富性和自然语言表达的多样性使得视觉场景描述成为一项充满挑战的任务,综述了现有视觉场景描述方法及其效果评价.首先,论述了视觉场景描述的定义?研究任务及方法分类,简要分析了视觉场景描述与多模态检索、跨模态学习、场景分类、视觉关系检测等相关技术的关系;然后分类讨论视觉场景描述的主要方法?模型及研究进展,归纳日渐增多的基准数据集;接下来,梳理客观评价视觉场景描述效果的主要指标和视觉场景描述技术面临的问题与挑战,最后讨论未来的应用前景.

深度视觉目标跟踪进展综述

视频目标跟踪是计算机视觉领域的一个重要研究课题.近年来,随着深度学习在视觉目标跟踪领域获得了巨大的成功,一系列优秀的深度跟踪算法涌现出来.本文回顾了近年来深度目标跟踪领域的进展.首先,我们详细讨论了近十年来跟踪领域数据集的发展趋势,这些数据集不仅全面地评估了算法性能同时为模型训练提供了极大的便利.其次,我们分类讨论了几大类经典的深度学习跟踪框架,包括深度相关滤波器跟踪、分类式网络跟踪、双路网络跟踪、基于梯度的深度跟踪算法以及基于Transformer的跟踪算法.最后,我们对全文内容进行总结,并指出未来的发展趋势.

列存储数据库中压缩位图索引技术

为提高压缩码的利用率,提出一种适用于列存储数据库的压缩位图索引技术。定义反转、合并等操作,将所有计算的输入值与输出值格式化为位向量形式。通过活跃度衡量索引中位向量的复杂度,并对压缩位向量进行直接计算,优化where子句和group by子句在查询执行过程中的数据提取。在SSB数据集上的实验结果证明,该技术能提高29.7%~38.9%的压缩位图索引性能。

基于背景相似性匹配的人脸识别

一般环境中在线视频的人脸识别问题成为当前的研究热点之一。尽管与特定环境中在线视频的人脸识别有着很大关联,但一般环境中在线视频的人脸识别问题有着自己的独特特点和算法要求。目前研究与实际应用间仍存在较大差距。在此,建立了一般环境中的人脸库;利用建立的人脸库,实现研究、比较各种人脸识别技术性能;提出了一种新的相似测量标准,即背景相似性匹配。

Parametric Evaluation of Video Motion Tracking Data Sets

Video tracking is a complex problem because the environment, in which video motion needs to be tracked, is widely varied based on the application and poses several constraints on the design and performance of the tracking system. Current datasets that are used to evaluate and compare video motion tracking algorithms use a cumulative performance measure without thoroughly analyzing the effect of these different constraints imposed by the environment. But it needs to analyze these constraints as parameters. The objective of this paper is to identify these parameters and define quantitative measures for these parameters to compare video datasets for motion tracking.

一种扩展VIFB的红外与可见光图像融合基准

红外与可见光图像融合是机器视觉的一个重要领域,在日常生活中应用广泛。近年来,虽然红外与可见光图像融合领域已有多种融合算法,但目前该领域还缺乏能够衡量多种融合算法性能的算法框架和融合基准。在简要概述了红外与可见光图像融合的最新进展后,提出了一种扩展VIFB的红外与可见光图像融合基准,该基准由56对图像、32种融合算法和16种评价指标组成。基于该融合基准进行了大量实验,用来测评所选取的融合算法的性能。通过定性和定量结果分析,确定了性能优良的图像融合算法,并对红外与可见光图像融合领域的未来前景进行了展望。

基于空域与频域关系建模的篡改文本图像检测

近年来,篡改文本图像在互联网的广泛传播为文本图像安全带来严重威胁。然而,相应的篡改文本检测(TTD,tampered text detection)方法却未得到充分的探索。TTD任务旨在定位图像中所有文本区域,同时根据纹理的真实性判断文本区域是否被篡改。与一般的文本检测任务不同,TTD任务需要进一步感知真实文本和篡改文本分类的细粒度信息。TTD任务有两个主要挑战:一方面,由于真实文本和篡改文本的纹理具有较高的相似性,仅在空域(RGB)进行纹理特征学习的篡改文本检测方法不能很好地区分两类文本;另一方面,由于检测真实文本和篡改文本的难度不同,检测模型无法平衡两类文本的学习过程,从而造成两类文本检测精度的不平衡问题。相较于空域特征,文本纹理在频域中的不连续性能够帮助网络鉴别文本实例的真伪,根据上述依据,提出基于空域和频域(RGB and frequency)关系建模的篡改文本检测方法。采用空域和频域特征提取器分别提取空域和频域特征,通过引入频域信息增强网络对篡改纹理的鉴别能力;使用全局空频域关系模块建模不同文本实例的纹理真实性关系,通过参考同幅图像中其他文本实例的空频域特征来辅助判断当前文本实例的真伪性,从而平衡真实和篡改文本检测难度,解决检测精度不平衡问题;提出一个票据篡改文本图像数据集(Tampered-SROIE)来验证上述篡改文本检测方法的有效性,该数据集包含986张图像(626张训练图像和360张测试图像)。该方法在Tampered-SROIE上的真实和篡改文本检测F值分别达到95.97%和96.80%,同时降低检测精度不平衡性1.13%。该方法从网络结构与检测策略的角度为篡改文本检测任务提供了新的解决方案,同时Tampered-SROIE为以后的篡改文本检测方法提供了评估基准。

面向人脸修复篡改检测的大规模数据集

目的 图像合成方法随着计算机视觉的不断发展和深度学习技术的逐渐成熟为人们的生活带来了丰富的体验。然而,用于传播虚假信息的恶意篡改图像可能对社会造成极大危害,使人们对数字内容在图像媒体中的真实性产生怀疑。面部编辑作为一种常用的图像篡改手段,通过修改面部的五官信息来伪造人脸。图像修复技术是面部编辑常用的手段之一,使用其进行面部伪造篡改同样为人们的生活带来了很大干扰。为了对此类篡改检测方法的相关研究提供数据支持,本文制作了面向人脸修复篡改检测的大规模数据集。方法 具体来说,本文选用了不同质量的源数据集(高质量的人脸图像数据集CelebA-HQ及低质量的人脸视频数据集FF++),通过图像分割方法将面部五官区域分割,最后使用两种基于深度网络的修复方法CTSDG(image inpainting via conditional texture and structure dual generation)和RFR(recurrent feature reasoning for image inpainting)以及一种传统修复方法 SC(struct completion),生成总数量达到60万幅的大规模修复图像数据集。结果 实验结果表明,由FF++数据集生成的图像在基准检测网络ResNet-50下的检测精度下降了15%,在Xception-Net网络下检测精度下降了5%。且不同面部部位的检测精度相差较大,其中眼睛部位的检测精度最低,检测精度为0.91。通过泛化性实验表明,同一源数据集生成的数据在不同部位的修复图像间存在一定的泛化性,而不同的源数据制作的数据集间几乎没有泛化性。因此,该数据集也可为修复图像之间的泛化性研究提供研究数据,可以在不同数据集、不同修复方式和不同面部部位生成的图像间进行修复图像的泛化性研究。结论 基于图像修复技术的篡改方式在一定程度上可以骗过篡改检测器,对于此类篡改方式的检测方法研究具有现实意义。提供的大型基于修复技术的人脸篡改数据集为该领域的研究提供了新的数据来源,丰富了数据多样性,为深入研究该类型的人脸篡改和检测方法提供了有力的基准。数据集开源地址https://pan.baidu.com/s/1-9HIBya9X-geNDe5zcJldw?pwd=thli。

Nphos:Database and Predictor of Protein N-phosphorylation

Protein N-phosphorylation is widely present in nature and participates in various biological processes.However,current knowledge on N-phosphorylation is extremely limited compared to that on O-phosphorylation.In this study,we collected 11,710 experimentally verified N-phosphosites of 7344 proteins from 39 species and subsequently constructed the database Nphos to share up-to-date information on protein N-phosphorylation.Upon these substantial data,we characterized the sequential and structural features of protein N-phosphorylation.Moreover,after comparing hundreds of learning models,we chose and optimized gradient boosting decision tree(GBDT)models to predict three types of human N-phosphorylation,achieving mean area under the receiver operating characteristic curve(AUC)values of 90.56%,91.24%,and 92.01%for pHis,pLys,and pArg,respectively.Meanwhile,we discovered 488,825 distinct N-phosphosites in the human proteome.The models were also deployed in Nphos for interactive N-phosphosite prediction.In summary,this work provides new insights and points for both flexible and focused investigations of N-phosphorylation.It will also facilitate a deeper and more systematic understanding of protein N-phosphorylation modification by providing a data and technical foundation.Nphos is freely available at http://www.bio-add.org/Nphos/and http://ppodd.org.cn/Nphos/.

面向无人机海岸带生态系统监测的语义分割基准数据集

目的海岸带生态系统的监测是维护生态平衡和可持续发展的重要任务。语义分割技术在遥感影像中的应用为海岸带生态系统的精准监测提供了有效手段。然而,目前该领域仍面临一个挑战,即缺乏公开的专门面向海岸带生态系统的语义分割基准数据集。方法受到自然环境和人为因素的共同影响,海岸带外部形态变化快,目前的卫星遥感和常规的人工测量、船载作业等监测方式已难以满足实时观测的要求。因此本文首先利用无人机对海岸带典型生态群落进行拍摄、采集和标注,构建了面向海岸带生态系统的基准数据集,名为OUC-UAV-SEG。接着,通过统计的方法,对OUC-UAV-SEG进行定量分析,并对数据集中存在的挑战进行了深入探讨。最后,使用典型视觉语义分割算法对OUC-UAV-SEG展开评估。结果对OUC-UAV-SEG数据集进行了基准测试,评估了当前可用的语义分割算法在该数据集上的性能,并揭示了这些算法的局限性。结论OUC-UAV-SEG数据集的建立为海岸带生态系统的监测提供了新的资源,有助于科学家、生态学家和决策者们更好地了解海岸带生态系统的现状和变化趋势,为海洋生态保护和管理提供更精确的信息。该数据集链接:https://github.com/OucCVLab/OUC-UAV-SEG。