行人再识别技术研究进展

行人再识别通过大时空范围内跨摄像机目标行人图像的检索与匹配,可实现人脸等生物特征失效情况下的行人关联,已成为智能视频监控系统的关键环节和支撑技术,并在智慧公安、智慧城市等国民经济建设中发挥了重要作用。近年行人再识别技术吸引了越来越多的关注,并取得了快速发展与进步。本文在对行人再识别技术进行简介的基础上,面向行人再识别的技术发展和落地应用需求与挑战,总结分析遮挡行人再识别、无监督行人再识别、虚拟数据生成、域泛化行人再识别、换装行人再识别、跨模态行人再识别和行人搜索等热点方向的前沿进展,归纳其发展现状和存在问题,最后对行人再识别技术的发展趋势进行展望。希望通过总结和分析,能够为研究人员开展行人再识别相关研究、推动行人再识别技术进步提供参考。     

多姿态图像生成的行人重识别算法研究

不同行人的高度相似性以及相同行人外观姿态的差异性,使得不同摄像头下的行人重识别面临严峻的挑战。生成对抗网络可以合成新的图像,被认为是解决行人姿态变化的主要技术手段。提出一种基于多姿态图像生成的行人重识别算法,利用生成对抗网络生成不同姿态的行人图像,通过归一化消除姿态的影响,从而大幅度提升行人重识别的整体性能。该行人重识别算法包括多姿态行人图像生成、不同姿态的行人特征提取与融合、距离度量和重排序三部分内容。在Market-1501数据集和DukeMTMC-ReID数据集上的实验证实了所提出算法的有效性,通过与state-of-the-art行人重识别方法比较,展示了多姿态图像生成方法在行人重识别任务中的优越性,同时表明生成行人图像的特征与原始图像的特征是相互补充的。     

多源数据行人重识别研究综述

行人重识别是近年来计算机视觉领域的热点问题, 经过多年的发展, 基于可见光图像的一般行人重识别技术已经趋近成熟. 然而, 目前的研究多基于一个相对理想的假设, 即行人图像都是在光照充足的条件下拍摄的高分辨率图像. 因此虽然大多数的研究都能取得较为满意的效果, 但在实际环境中并不适用. 多源数据行人重识别即利用多种行人信息进行行人匹配的问题. 除了需要解决一般行人重识别所面临的问题外, 多源数据行人重识别技术还需要解决不同类型行人信息与一般行人图片相互匹配时的差异问题, 如低分辨率图像、红外图像、深度图像、文本信息和素描图像等. 因此, 与一般行人重识别方法相比, 多源数据行人重识别研究更具实用性, 同时也更具有挑战性. 本文首先介绍了一般行人重识别的发展现状和所面临的问题, 然后比较了多源数据行人重识别与一般行人重识别的区别, 并根据不同数据类型总结了5 类多源数据行人重识别问题, 分别从方法、数据集两个方面对现有工作做了归纳和分析. 与一般行人重识别技术相比, 多源数据行人重识别的优点是可以充分利用各类数据学习跨模态和类型的特征转换. 最后, 本文讨论了多源数据行人重识别未来的发展.     

基于混合方法的中文人名识别研究

当前中文人名识别的研究主要针对中国人名,而对日本人名及音译人名的专门研究相对较少,识别效果也亟待提高。提出利用CRRM方法进行中、日及音译人名同步识别。该方法基于CRF（Conditional Random Fields）并结合了上下文规则及人名可信度模型。此外,利用局部统计算法对边界识别错误的人名进行修正,并利用扩散操作召回未被识别的人名。实验结果表明,中、日、音译人名识别的F值均高于90%,提出的方法可以取得较好的识别效果。     

基于深度学习的行人再识别技术研究综述

行人再识别（Re-id）作为智能视频监控技术之一,其目的是在不同的摄像机视图中检索出指定身份的行人,因此该项技术对维护社会治安稳定具有重大研究意义。针对传统的手工特征方法难以应对行人Re-id任务中复杂的摄像机环境的问题,大量基于深度学习的行人Re-id方法被提出,极大地推动了行人Re-id技术的发展。为了深入了解基于深度学习的行人Re-id技术,整理和分析了大量相关文献,首先从图像、视频、跨模态这3个方面展开综述性介绍,将图像行人Re-id技术分为有监督和无监督两大类并分别进行概括;然后列举了部分相关数据集,并对近年来在图像和视频数据集上的一些算法进行性能的比较与分析;最后总结了行人Re-id技术的发展难点,并深入讨论了该技术未来可能的研究方向。     

基于深度学习的行人再识别技术研究综述

行人再识别（Re-id）作为智能视频监控技术之一，其目的是在不同的摄像机视图中检索出指定身份的行人，因此该项技术对维护社会治安稳定具有重大研究意义。针对传统的手工特征方法难以应对行人Re-id任务中复杂的摄像机环境的问题，大量基于深度学习的行人Re-id方法被提出，极大地推动了行人Re-id技术的发展。为了深入了解基于深度学习的行人Re-id技术，整理和分析了大量相关文献，首先从图像、视频、跨模态这3个方面展开综述性介绍，将图像行人Re-id技术分为有监督和无监督两大类并分别进行概括；然后列举了部分相关数据集，并对近年来在图像和视频数据集上的一些算法进行性能的比较与分析；最后总结了行人Re-id技术的发展难点，并深入讨论了该技术未来可能的研究方向。     

Identification of a specific person using color,height, and gait features for a person following robot

This paper describes a person identification method for mobile service robots using image and range data. Person identification is a necessary function in order for mobile service robots to locate the target person for those services. Among various sensory features, image-based appearance features have often been used for person identification. They are, however, not effective in severe illumination environments such as a strong backlight. Therefore, we use two illumination-independent features, height and gait, in addition to appearance features for a more robust identification. To this end, we have developed a new method of extracting the gait feature (step length and speed), based on a maximum likelihood estimation of supporting leg positions in accumulated range data. We combine these features and use an online boosting approach to create the specific person classifier. It allows the robot to identify the specific person robustly even in a severe illumination environment. We tested our multi-feature person identification method, combined with a range data-based person tracker, in a specific person following scenario to demonstrate the effectiveness of this method.     

联合均等采样随机擦除和全局时间特征池化的视频行人重识别方法

针对为解决视频监控中遮挡、背景物干扰,以及行人外观、姿势相似性等因素导致的视频行人重识别准确率较低的问题,提出了联合均等采样随机擦除和全局时间特征池化的视频行人重识别方法。首先针对目标行人被干扰或部分遮挡的情况,采用了均等采样随机擦除（ESE）的数据增强方法来有效地缓解遮挡,提高模型的泛化能力,更准确地匹配行人;其次为了进一步提高视频行人重识别的精度,学习更有判别力的特征表示,使用三维卷积神经网络（3DCNN）提取时空特征,并在网络输出行人特征表示前加上全局时间特征池化层（GTFP）,这样既能获取上下文的空间信息,又能细化帧与帧之间的时序信息。通过在MARS、DukeMTMC-VideoReID 和PRID-2011三个公共视频数据集上的大量实验,证明所提出的联合均等采样随机擦除和全局时间特征池化的方法,相较于目前一些先进的视频行人重识别方法,具有一定的竞争力。     

基于双目标定的独居老人摔倒检测算法

独居老人摔倒已成为一个备受关注的问题.为快速有效获取摔倒信息从而使老人得到及时救助,提出一种基于双目标定的独居老人摔倒检测算法.该算法通过色彩不变性分割前景目标（老人）,采用双目视觉标定计算人体在三维坐标中高度作为特征信息,能够有效区分易混淆动作,防止误判,提高检测准确率.实验结果表明：该算法易于实现,具有较好的鲁棒性和实时性.     

An information fusion framework for person localization via body pose in spectator crowds

Person localization or segmentation in low resolution crowded scenes is important for person tracking and recognition, action detection and anomaly identification. Due to occlusion and lack of inter-person space, person localization becomes a difficult task. In this work, we propose a novel information fusion framework to integrate a Deep Head Detector and a body pose detector. A more accurate body pose showing limb positions will result in more accurate person localization. We propose a novel Deep Head Detector (DHD) to detect person heads in crowds. The proposed DHD is a fully convolutional neural network and it has shown improved head detection performance in crowds. We modify Deformable Parts Model (DPM) pose detector to detect multiple upper body poses in crowds. We efficiently fuse the information obtained by the proposed DHD and the modified DPM to obtain a more accurate person pose detector. The proposed framework is named as Fusion DPM (FDPM) and it has exhibited improved body pose detection performance on spectator crowds. The detected body poses are then used for more accurate person localization by segmenting each person in the crowd.     

Facial Recognition System for Suspect Identification Using a Surveillance Camera

Nowadays, finding and Tracking a person in the world of technology is becoming a necessary task for various security purposes. Since the advent of technology, the development in the field of Facial Recognition plays an important role and has been exponentially increasing in today’s world. In this, a model is proposed for facial recognition to identify and alert the system when a person in search has been found at a specific location under the surveillance of a CCTV camera. The CCTV cameras are connected to a centralized server to which the live streaming feed is uploaded by cameras at each location. The server contains a database of all persons to be found. Based on the video feed from each camera, if a particular person in search is found in a certain feed, then the location of that person will be tracked and also a signal is passed to the system responsible. This model is based on image processing concepts to match live images with the existing trained images of the person in search. Since this model recognizes a person based on the first and foremost primary unique feature of a human, that is, only the person’s face image is required and will be found to be stored in the database. Hence the task of finding a person reduces to the task of detecting human faces in the video feed and matching with the existing images from the database.     

基于迁移学习的行人再识别

行人再识别指的是在无重叠的多摄像机监控视频中,匹配不同摄像机中的行人目 标。提出了一种基于迁移学习的行人再识别方法。在训练阶段,针对现有的基于深度卷积神经 网络的图像识别模型进行参数微调,将网络模型迁移学习至行人再识别模型。测试阶段,利用 学习好的网络模型提取行人图像的特征,再采用余弦距离来描述行人对之间的相似度。在 CUHK03、Market-1501 和 DukeMTMC-reID 3 个数据集上进行了深入的实验分析,实验结果表 明该方法取得了较高的累积匹配得分,特别是第 1 匹配率远远超过了非深度学习的方法,与其 他基于深度学习的行人再识别方法相比,准确率也有所提升。     

基于社会网络的人名检索结果重名消解

人物重名现象十分普遍,搜索引擎的人名检索结果通常是多个同名人物相关网页的混合.该文依据同名的不同人物具有不同的社会网络的思想,利用检索结果中共现的人名发现并拓展检索人物相关的潜在社会网络,结合图的谱分割算法和模块度指标进行社会网络的自动聚类,在此基础上实现人名检索结果的重名消解.在人工标注的中文人名语料上进行实验,整体性能达到较好水平,图聚类算法能帮助连通社会网络的进一步划分,从而提高消解效果.     

Visual person identification using a distance-dependent appearance model for a person following robot

This paper describes a person identifcation method for a mobile robot which performs specifc person following under dynamic complicated environments like a school canteen where many persons exist.We propose a distance-dependent appearance model which is based on scale-invariant feature transform(SIFT) feature.SIFT is a powerful image feature that is invariant to scale and rotation in the image plane and also robust to changes of lighting condition.However,the feature is weak against afne transformations and the identifcation power will thus be degraded when the pose of a person changes largely.We therefore use a set of images taken from various directions to cope with pose changes.Moreover,the number of SIFT feature matches between the model and an input image will decrease as the person becomes farther away from the camera.Therefore,we also use a distance-dependent threshold.The person following experiment was conducted using an actual mobile robot,and the quality assessment of person identifcation was performed.     

基于MVC模式的人才测评系统的分析与设计

文章对人才测评系统进行了功能需求上的分析，在基于MVC模式和．NET框架下，建立了一种人才素质模型，并给出了一种人才测评系统的设计开发方案。     

A benchmark for clothes variation in person re-identification

Person re-identification (re-ID) has drawn attention significantly in the computer vision society due to its application and research significance. It aims to retrieve a person of interest across different camera views. However, there are still several factors that hinder the applications of person re-ID. In fact, most common data sets either assume that pedestrians do not change their clothing across different camera views or are taken under constrained environments. Those constraints simplify the person re-ID task and contribute to early development of person re-ID, yet a person has a great possibility to change clothes in real life. To facilitate the research toward conquering those issues, this paper mainly introduces a new benchmark data set for person re-identification. To the best of our knowledge, this data set is currently the most diverse for person re-identification. It contains 107 persons with 9,738 images, captured in 15 indoor/outdoor scenes from September 2019 to December 2019, varying according to viewpoints, lighting, resolutions, human pose, seasons, backgrounds, and clothes especially. We hope that this benchmark data set will encourage further research on person re-identification with clothes variation. Moreover, we also perform extensive analyses on this data set using several state-of-the-art methods. Our dataset is available at https://github.com/nkicsl/NKUP-dataset .     

Specific person tracking using 3D LIDAR and ESPAR antenna for mobile service robots

The ability of detecting and following a specific person is indispensable for mobile service robots. Many image-based methods have been proposed for person detection and identification; however, they are sometimes vulnerable to illumination changes. This paper therefore proposes a novel approach to the problem, namely, using 3D LIDARs for person detection and identification and a directivity-controllable antenna (called ESPAR antenna) for localizing a specific person even under long-term occlusion and/or out-of-view situations. A sensor fusion framework, combined with an adaptive state-based strategy switching, has also been developed for achieving a reliable person following. Experimental results in actual outdoor environments show the effectiveness of the proposed framework.     

行人再识别技术综述

行人再识别指的是判断不同摄像头下出现的行人是否属于同一行人, 可以看作是图像检索的子问题, 可以广泛应用于智能视频监控、安保、刑侦等领域.由于行人图像的分辨率变化大、拍摄角度不统一、光照条件差、环境变化大、行人姿态不断变化等原因, 使得行人再识别成为目前计算机视觉领域一个既具有研究价值又极具挑战性的研究热点和难点问题.早期的行人再识别方法大多基于人工设计特征, 在小规模数据集上开展研究.近年来, 大规模行人再识别数据集不断推出, 以及深度学习技术的迅猛发展, 为行人再识别技术的发展带来了新的契机.本文对行人再识别的发展历史、研究现状以及典型方法进行梳理和总结.首先阐述了行人再识别的基本研究框架, 然后分别针对行人再识别的两个关键技术(特征表达和相似性度量), 进行了归纳总结, 重点介绍了目前发展迅猛的深度学习技术在行人再识别中的应用.另外, 本文对行人再识别中代表性的数据集以及在各个数据集上可以取得优异性能的方法进行了分析和比较.最后对行人再识别技术的未来发展趋势进行了展望.     

基于SVM的人物实体全局模式动态构建

针对结构化网页中人物实体全局模式构建问题,提出了基于SVM的动态构建方法。构建分为两个阶段,第一个阶段是来自同一数据源的人物实体结构化实例到人物实体局部模式的转化,第二个阶段是利用SVM分类器完成人物实体局部模式到人物实体全局模式的映射。本方法能适应数据源的不断变化,保证了全局模式的完整性。通过实验,验证了构建算法的有效性和可行性,并对随着结构化网页不断增多时全局模式的稳定性进行了考察。