视频群体行为识别综述

群体行为识别是指给定一个包含多人场景的视频,模型需要识别出视频中多个人物正在共同完成的群体行为.群体行为识别是视频理解中的一个重要问题,可以被应用在运动比赛视频分析、监控视频识别、社交行为理解等现实场景中.多人场景视频较为复杂,时间和空间上的信息十分丰富,对模型提取关键信息的能力要求更高.模型只有高效地建模场景中的层次化关系,并为人物群体提取有区分性的时空特征,才能准确地识别出群体行为.由于其广泛的应用需求,群体行为识别问题受到了研究人员的广泛关注.对近几年来群体行为识别问题上的大量研究工作进行了深入分析,总结出了群体行为识别研究所面临的主要挑战,系统地归纳出了6种类型的群体行为识别方法,包含传统非深度学习识别方法以及基于深度学习技术的识别方法,并对未来研究的可能方向进行了展望.     

Independent shape component-based human activity recognition via Hidden Markov Model

In proactive computing, human activity recognition from image sequences is an active research area. In this paper, a novel human activity recognition method is proposed, which utilizes Independent Component Analysis (ICA) for activity shape information extraction from image sequences and Hidden Markov Model (HMM) for recognition. Various human activities are represented by shape feature vectors from the sequence of activity shape images via ICA. Based on these features, each HMM is trained and activity recognition is achieved by the trained HMMs of different activities. Our recognition performance has been compared to the conventional method where Principal Component Analysis (PCA) is typically used to derive activity shape features. Our results show that superior recognition is achieved with the proposed method especially for activities (e.g., skipping) that cannot be easily recognized by the conventional method. Furthermore, by employing Linear Discriminant Analysis (LDA) on IC features, the recognition results further improved significantly in the recognition performance.     

基于智能手机传感器的人体活动识别

人体活动识别是上下文感知系统及其应用中一个具有挑战性的研究问题。目前,关于人体活动识别的研究主要使用一些基于监督学习或半监督学习的统计方法来构建识别模型。然而,考虑到识别活动类型本身具有的复杂性和多样性,当前的人体活动识别系统不能取得较好的识别效果。针对这一问题,通过智能手机的三维加速度和陀螺仪传感器信息来提取人体活动的特征向量,选择四种典型的统计学习方法（分别是K-近邻算法、支持向量机、朴素贝叶斯网络以及基于朴素贝叶斯网络的AdaBoost算法）分别创建人体活动的识别模型,最后通过模型决策得到最优的人体活动识别模型。实验结果表明,通过模型决策选择的识别模型对人体活动识别准确率达到92%,取得很好的识别效果。     

Dynamic sensor data segmentation for real-time knowledge-driven activity recognition

Approaches and algorithms for activity recognition have recently made substantial progress due to advancements in pervasive and mobile computing, smart environments and ambient assisted living. Nevertheless, it is still difficult to achieve real-time continuous activity recognition as sensor data segmentation remains a challenge. This paper presents a novel approach to real-time sensor data segmentation for continuous activity recognition. Central to the approach is a dynamic segmentation model, based on the notion of varied time windows, which can shrink and expand the segmentation window size by using temporal information of sensor data and activities as well as the state of activity recognition. The paper first analyzes the characteristics of activities of daily living from which the segmentation model that is applicable to a wide range of activity recognition scenarios is motivated and developed. It then describes the working mechanism and relevant algorithms of the model in the context of knowledge-driven activity recognition based on ontologies. The presented approach has been implemented in a prototype system and evaluated in a number of experiments. Results have shown average recognition accuracy above 83% in all experiments for real time activity recognition, which proves the approach and the underlying model.     

基于神经网络的人体动作识别方法

人体动作识别一直是计算机视觉领域的研究重点。为了提高人体动作识别的准确度,本文提出一种基于神经网络的加权识别方法。首先利用ViBe算法提取人体运动前景,计算前景重心,然后将轮廓重心距作傅里叶变换获得傅里叶描述子,最后利用本文提出的基于神经网络的加权识别方法进行分类。实验结果表明,本文方法的识别率在89%以上。     

基于CNN-LSTM的活动识别方法

在识别活动时,传统的循环神经网络RNN识别方法不考虑传感器活动数据之间依赖性强的问题,导致识别准确率降低。为了提高识别准确率,解决活动数据依赖性强的问题,用长短期记忆网络LSTM进行活动识别,LSTM在考虑当前点输入的同时考虑先前点的输出,能够保持数据之间的强依赖性。但是,LSTM在处理传感器活动数据的特征提取方面时间效率不高,而卷积神经网络CNN能共享卷积核,且可以从杂乱无章的数据中提取出明显特征向量。提出一种基于CNN-LSTM的活动识别方法CLAR,利用CNN能够很好地提取出活动序列数据中的特征向量,并将提取出的特征向量作为LSTM的输入,利用LSTM门限之间的相互作用进行活动识别,使得依赖性很强的活动数据成为活动识别的优势,进而提高活动识别的准确率和时间效率。实验表明,CLAR方法的识别准确率比单一神经网络活动识别方法的准确率提高了9%,时间平均缩短了10%。     

基于联合特征和XGBoost的活动语义识别方法

针对以往活动语义识别研究单纯提取时间维度上的序列特征以及周期特征、缺乏对空间信息的深度挖掘等问题，提出一种基于联合特征和极限梯度提升（XGBoost）的活动语义识别方法。首先，挖掘时间信息中的活动周期性特征和空间信息中的经纬度特征；然后，使用经纬度信息通过具有噪声的基于密度的聚类（DBSCAN）算法提取空间区域热度特征，将这些特征组成特征向量来刻画用户活动语义；最后，采用集成学习方法中的XGBoost算法建立活动语义识别模型。在FourSquare的两个公共签到数据集上，基于联合特征的模型比基于时间特征的模型在识别准确率上提高了28个百分点，与上下文感知混合（CAH）方法和时空活动偏好（STAP）方法对比，所提方法的识别准确率分别提高了30个百分点和5个百分点。实验结果表明所提方法与对比方法相比在活动语义识别问题上更加准确有效。     

面向人-物交互的视觉识别方法研究

从物体识别、动作识别和人-物交互行为识别三者关系角度对人-物交互的视觉识别方法进行了研究。将三者关系建模成统一的人-物交互识别的模型,通过该模型对以动作作为上下境进行物体识别,以物体作为上下境进行动作识别、交互行为识别、基于图模型的对象识别以及由交互行为进行动作和物体识别5个方面进行了详细研究,最后对面向人-物交互识别方法进行了比较。     

基于联合特征和XGBoost的活动语义识别方法

针对以往活动语义识别研究单纯提取时间维度上的序列特征以及周期特征、缺乏对空间信息的深度挖掘等问题,提出一种基于联合特征和极限梯度提升（XGBoost）的活动语义识别方法。首先,挖掘时间信息中的活动周期性特征和空间信息中的经纬度特征;然后,使用经纬度信息通过具有噪声的基于密度的聚类（DBSCAN）算法提取空间区域热度特征,将这些特征组成特征向量来刻画用户活动语义;最后,采用集成学习方法中的XGBoost算法建立活动语义识别模型。在FourSquare的两个公共签到数据集上,基于联合特征的模型比基于时间特征的模型在识别准确率上提高了28个百分点,与上下文感知混合（CAH）方法和时空活动偏好（STAP）方法对比,所提方法的识别准确率分别提高了30个百分点和5个百分点。实验结果表明所提方法与对比方法相比在活动语义识别问题上更加准确有效。     

Extreme learning machine-based device displacement free activity recognition model

Activity recognition based on mobile device is an important aspect in developing human-centric pervasive applications like gaming, industrial maintenance and health monitoring. However, the data distribution of accelerometer is heavily affected by varying device locations and orientations, which will degrade the performance of recognition model. To solve this problem, we propose a fast, robust and device displacement free activity recognition model in this paper, which integrates principal component analysis (PCA) and extreme learning machine (ELM) to realize location-adaptive activity recognition. On the one hand, PCA is employed to reduce the dimensionality of feature space and extract robust features for recognition. On the other hand, in the online phase ELM is applied to classify the activity and adapt the recognition model to new device locations based on high confident recognition results in real time. Experimental results show that, with robust features and fast adaptation capability, the proposed model can adapt the classifier to new device locations quickly and obtain good recognition performance.     

一种基于多分类器融合的人体运动行为识别模型

为了提高基于智能设备的人体日常行为识别的准确率,针对不同智能设备内置加速度传感器获取的三轴加速度信息,提出了一种基于多分类器融合的行为识别MCF(Multiple Classifier Fusion)模型。针对5种日常行为(静止、散步、跑步、上楼及下楼),优选出与每种行为相关度高的特征集,用于训练对每种行为识别效果最佳的5个基分类器,并采用一个融合器对5个基分类器的输出进行融合处理,得到最终行为识别结果。该模型对这5种行为的平均识别准确率和可信度分别达到96.84%和97.41%,能有效进行用户行为识别。     

Robust least squares twin support vector machine for human activity recognition

Human activity recognition is an active area of research in Computer Vision. One of the challenges of activity recognition system is the presence of noise between related activity classes along with high training and testing time complexity of the system. In this paper, we address these problems by introducing a Robust Least Squares Twin Support Vector Machine (RLS-TWSVM) algorithm. RLS-TWSVM handles the heteroscedastic noise and outliers present in activity recognition framework. Incremental RLS-TWSVM is proposed to speed up the training phase. Further, we introduce the hierarchical approach with RLS-TWSVM to deal with multi-category activity recognition problem. Computational comparisons of our proposed approach on four well-known activity recognition datasets along with real world machine learning benchmark datasets have been carried out. Experimental results show that our method is not only fast but, yields significantly better generalization performance and is robust in order to handle heteroscedastic noise and outliers.     

基于块稀疏贝叶斯学习的人体运动模式识别

在人体运动模式识别中, 传统稀疏表示分类算法未考虑待测试样本相应稀疏系数向量内在块结构相关性信息,影响了算法识别性能。为此,提出一种基于块稀疏模型的人体运动模式识别方法。该方法充分利用人体运动模式内在块稀疏结构,将人体运动模式识别问题转化为稀疏表示问题,采用块稀疏贝叶斯学习算法,求解基于样本训练集优化稀疏表示待测样本的稀疏系数, 并根据稀疏系数重构残差判定待识别动作类别,能有效提高人体运动模式识别率。选用包含多类别人体动作行为模式的USC-HAD数据库对所提算法性能进行了验证。实验结果表明,所提算法能够有效捕获不同运动模式内在差异信息,平均动作识别率达到97.86%,比传统动作识别方法平均提高近5%,有效提高了动作识别准确率。     

智能家居中的居民日常行为识别综述

针对智能家居中居民日常行为识别进行了综述。介绍了基于非入侵式传感器的智能家居环境,分析了智能家居中日常行为识别研究的意义;概述了日常行为识别的基本流程;根据日常行为识别的基本流程,归纳了目前日常行为识别研究领域中传感器事件流分割、日常行为特征的选择和计算以及日常行为识别方法三个关键问题的研究现状;指出了未来智能家居中日常行为识别的研究方向。     

Enhancing activity recognition using CPD-based activity segmentation

Segmenting behavior-based sensor data and recognizing the activity that the data represents are vital steps in all applications of human activity learning such as health monitoring, security, and intervention. In this paper, we enhance activity recognition by identifying activity transitions. To accomplish this goal, we introduce a change point detection-based activity segmentation model which partitions behavior-driven sensor data into non-overlapping activities in real time. In addition to providing valuable activity information, activity segmentation also can be used to improve the performance of activity recognition. We evaluate our proposed segmentation-enhanced activity recognition method on data collected from 29 smart homes. Results of this analysis indicate that the method not only provides useful information about activity boundaries and transitions between activities but also increases recognition accuracy by 7.59% and f measure by 6.69% in comparison with the traditional window-based methods.     

Transfer learning for activity recognition: a survey

Many intelligent systems that focus on the needs of a human require information about the activities being performed by the human. At the core of this capability is activity recognition, which is a challenging and well-researched problem. Activity recognition algorithms require substantial amounts of labeled training data yet need to perform well under very diverse circumstances. As a result, researchers have been designing methods to identify and utilize subtle connections between activity recognition datasets, or to perform transfer-based activity recognition. In this paper, we survey the literature to highlight recent advances in transfer learning for activity recognition. We characterize existing approaches to transfer-based activity recognition by sensor modality, by differences between source and target environments, by data availability, and by type of information that is transferred. Finally, we present some grand challenges for the community to consider as this field is further developed.     

On the use of ensemble of classifiers for accelerometer-based activity recognition

Activity recognition aims to detect the physical activities such as walking, sitting, and jogging performed by humans. With the widespread adoption and usage of mobile devices in daily life, several advanced applications of activity recognition were implemented and distributed all over the world. In this study, we explored the power of ensemble of classifiers approach for accelerometer-based activity recognition and built a novel activity prediction model based on machine learning classifiers. Our approach utilizes from J48 decision tree, Multi-Layer Perceptrons (MLP) and Logistic Regression techniques and combines these classifiers with the average of probabilities combination rule. Publicly available activity recognition dataset known as WISDM (Wireless Sensor Data Mining) which includes information from thirty six users was used during the experiments. According to the experimental results, our model provides better performance than MLP-based recognition approach suggested in previous study. These results strongly suggest researchers applying ensemble of classifiers approach for activity recognition problem.     

Discrete techniques applied to low-energy mobile human activity recognition. A new approach

Human activity recognition systems are currently implemented by hundreds of applications and, in recent years, several technology manufacturers have introduced new wearable devices for this purpose. Battery consumption constitutes a critical point in these systems since most are provided with a rechargeable battery. In this paper, by using discrete techniques based on the Ameva algorithm, an innovative approach for human activity recognition systems on mobile devices is presented. Furthermore, unlike other systems in current use, this proposal enables recognition of high granularity activities by using accelerometer sensors. Hence, the accuracy of activity recognition systems can be increased without sacrificing efficiency. A comparative is carried out between the proposed approach and an approach based on the well-known neural networks.     

基于多级深度网络架构的群体行为分析模型研究

群体行为的多层次深度分析是行为识别领域亟待解决的重要问题。在深度神经网络研究的基础上,提出了群体行为识别的层级性分析模型。基于调控网络的迁移学习,实现了行为群体中多人体的时序一致性检测;通过融合时空特征学习,完成了群体行为中时长无约束的个体行为识别;通过场景中个体行为类别、交互场景上下文信息的融合,实现了对群体行为稳定有效的识别。在公用数据集上进行的大量实验表明,与现有方法相比,该模型在群体行为分析识别方面具有良好的效果。