Robust object tracking via online dynamic spatial bias appearance models

This paper presents a robust object tracking method via a spatial bias appearance model learned dynamically in video. Motivated by the attention shifting among local regions of a human vision system during object tracking, we propose to partition an object into regions with different confidences and track the object using a dynamic spatial bias appearance model (DSBAM) estimated from region confidences. The confidence of a region is estimated to re ect the discriminative power of the region in a feature space, and the probability of occlusion. We propose a novel hierarchical Monte Carlo (HAMC) algorithm to learn region confidences dynamically in every frame. The algorithm consists of two levels of Monte Carlo processes implemented using two particle filtering procedures at each level and can efficiently extract high confidence regions through video frames by exploiting the temporal consistency of region confidences. A dynamic spatial bias map is then generated from the high confidence regions, and is employed to adapt the appearance model of the object and to guide a tracking algorithm in searching for correspondences in adjacent frames of video images. We demonstrate feasibility of the proposed method in video surveillance applications. The proposed method can be combined with many other existing tracking systems to enhance the robustness of these systems.     

Hybrid generative–discriminative hash tracking with spatio-temporal contextual cues

Visual object tracking is of a great application value in video monitoring systems. Recent work on video tracking has taken into account spatial relationship between the targeted object and its background. In this paper, the spatial relationship is combined with the temporal relationship between features on different video frames so that a real-time tracker is designed based on a hash algorithm with spatio-temporal cues. Different from most of the existing work on video tracking, which is regarded as a mechanism for image matching or image classification alone, we propose a hierarchical framework and conduct both matching and classification tasks to generate a coarse-to-fine tracking system. We develop a generative model under a modified particle filter with hash fingerprints for the coarse matching by the maximum a posteriori and a discriminative model for the fine classification by maximizing a confidence map based on a context model. The confidence map reveals the spatio-temporal dynamics of the target. Because hash fingerprint is merely a binary vector and the modified particle filter uses only a small number of particles, our tracker has a low computation cost. By conducting experiments on eight challenging video sequences from a public benchmark, we demonstrate that our tracker outperforms eight state-of-the-art trackers in terms of both accuracy and speed.

     

利用时空特征编码的单目标跟踪网络

目的 随着深度神经网络的出现,视觉跟踪快速发展,视觉跟踪任务中的视频时空特性,尤其是时序外观一致性(temporal appearance consistency)具有巨大探索空间。本文提出一种新颖简单实用的跟踪算法——时间感知网络(temporal-aware network, TAN),从视频角度出发,对序列的时间特征和空间特征同时编码。方法 TAN内部嵌入了一个新的时间聚合模块(temporal aggregation module, TAM)用来交换和融合多个历史帧的信息,无需任何模型更新策略也能适应目标的外观变化,如形变、旋转等。为了构建简单实用的跟踪算法框架,设计了一种目标估计策略,通过检测目标的4个角点,由对角构成两组候选框,结合目标框选择策略确定最终目标位置,能够有效应对遮挡等困难。通过离线训练,在没有任何模型更新的情况下,本文提出的跟踪器TAN通过完全前向推理(fully feed-forward)实现跟踪。结果 在OTB(online object tracking:a benchmark)50、OTB100、TrackingNet、LaSOT(a high-qua...     

基于视觉的目标检测与跟踪综述

基于视觉的目标检测与跟踪是图像处理、计算机视觉、模式识别等众多学科的交叉研究课题,在视频监控、虚拟现实、人机交互、自主导航等领域,具有重要的理论研究意义和实际应用价值.本文对目标检测与跟踪的发展历史、研究现状以及典型方法给出了较为全面的梳理和总结.首先,根据所处理的数据对象的不同,将目标检测分为基于背景建模和基于前景建模的方法,并分别对背景建模与特征表达方法进行了归纳总结.其次,根据跟踪过程有无目标检测的参与,将跟踪方法分为生成式与判别式,对基于统计的表观建模方法进行了归纳总结.然后,对典型算法的优缺点进行了梳理与分析,并给出了其在标准数据集上的性能对比.最后,总结了该领域待解决的难点问题,对其未来的发展趋势进行了展望.     

Robust visual tracking via discriminative appearance model based on sparse coding

In this paper, we formulate visual tracking as a binary classification problem using a discriminative appearance model. To enhance the discriminative strength of the classifier in separating the object from the background, an over-complete dictionary containing structure information of both object and background is constructed which is used to encode the local patches inside the object region with sparsity constraint. These local sparse codes are then aggregated for object representation, and a classifier is learned to discriminate the target from the background. The candidate sample with largest classification score is considered as the tracking result. Different from recent sparsity-based tracking approaches that update the dictionary using a holistic template, we introduce a selective update strategy based on local image patches which alleviates the visual drift problem, especially when severe occlusion occurs. Experiments on challenging video sequences demonstrate that the proposed tracking algorithm performs favorably against several state-of-the-art methods.     

基于深度学习的多目标跟踪关联模型设计

近年来, 深度学习在计算机视觉领域的应用取得了突破性进展, 但基于深度学习的视频多目标跟踪(Multiple object tracking, MOT)研究却相对甚少, 而鲁棒的关联模型设计是基于检测的多目标跟踪方法的核心.本文提出一种基于深度神经网络和度量学习的关联模型:采用行人再识别(Person re-identification, Re-ID)领域中广泛使用的度量学习技术和卷积神经网络(Convolutional neural networks, CNNs)设计目标外观模型, 即利用三元组损失函数设计一个三通道卷积神经网络, 提取更具判别性的外观特征构建目标外观相似度; 再结合运动模型计算轨迹片间的关联概率.在关联策略上, 采用匈牙利算法, 首先以逐帧关联方式得到短小可靠的轨迹片集合, 再通过自适应时间滑动窗机制多级关联, 输出各目标最终轨迹.在2DMOT2015、MOT16公开数据集上的实验结果证明了所提方法的有效性, 与当前一些主流算法相比较, 本文方法取得了相当或者领先的跟踪效果.     

基于判别性局部联合稀疏模型的多任务跟踪

目标表观建模是基于稀疏表示的跟踪方法的研究重点, 针对这一问题, 提出一种基于判别性局部联合稀疏表示的目标表观模型, 并在粒子滤波框架下提出一种基于该模型的多任务跟踪方法(Discriminative local joint sparse appearance model based multitask tracking method, DLJSM).该模型为目标区域内的局部图像分别构建具有判别性的字典, 从而将判别信息引入到局部稀疏模型中, 并对所有局部图像进行联合稀疏编码以增强结构性.在跟踪过程中, 首先对目标表观建立上述模型; 其次根据目标表观变化的连续性对采样粒子进行初始筛选以提高算法的效率; 然后求解剩余候选目标状态的联合稀疏编码, 并定义相似性函数衡量候选状态与目标模型之间的相似性; 最后根据最大后验概率估计目标当前的状态.此外, 为了避免模型频繁更新而引入累积误差, 本文采用每5帧判断一次的方法, 并在更新时保留首帧信息以减少模型漂移.实验测试结果表明DLJSM方法在目标表观发生巨大变化的情况下仍然能够稳定准确地跟踪目标, 与当前最流行的13种跟踪方法的对比结果验证了DLJSM方法的高效性.     

基于外观模型的目标跟踪算法研究进展

基于视觉的目标跟踪是模式识别、计算机视觉、机器学习等多个学科的交叉研究课题,在视频监控、视频压缩编码、视频检索、智能交通等领域有着十分广泛的应用。为了使国内外同行对基于外观模型的目标跟踪方法有一个较为全面的了解,对其进行了系统总结。在介绍跟踪算法原理的基础上,重点阐述了两大类基于外观模型的目标跟踪方法:产生式方法和判别式方法,深入讨论了其中的典型算法和研究成果,并对这些算法在公开数据集上的测试结果进行了分析比较,最后展望了该领域未来的发展方向。     

基于特征选择与时间一致性稀疏外观模型的目标追踪算法

为了更有效利用追踪目标的判别特征信息,提高目标追踪的精度和鲁棒性,在粒子滤波追踪框架下提出基于特征选择与时间一致性稀疏外观模型的目标追踪算法.首先,采集目标的正负模板和候选目标,根据特征选择模型对正负模板和候选目标进行特征选择,去除多余的干扰信息,得到关键的特征信息.然后,利用正负模板和候选目标的特征建立多任务稀疏表示模型,引入时间一致性正则项,促进更多的候选目标与先前帧的追踪结果具有稀疏表示的相似性.最后,求解多任务稀疏表示模型,得到判别稀疏相似图,获取每个候选目标的判别分,根据目标追踪结果更新正负模板.实验表明,即使在复杂的环境下,文中算法仍然比其它一些追踪算法具有更高的准确性.     

Learning deep features for online person tracking using non-overlapping cameras: A survey

Target-agnostic person tracking and re-identification across multiple non-overlapping cameras is an open vision problem. It is the task of maintaining the correct identity of people at different time instances and possibly different cameras. This study focuses on existing algorithms that facilitate online person tracking by using discriminative spatio-temporal features from video data, and presents the open issues and future research directions. The initial take on the problem introduces person tracking as a pure association problem, where the influence of human appearance, biometric and location information on re-identification are addressed explicitly. These constraints are modeled and used to understand and associate detections in real world environments. Next, a spatio-temporal model using LSTM networks for propagating associations and recovering from errors by taking advantage of the spatial and temporal information in videos is described. The spatio-temporal context indicates a way for discriminative appearance learning. The novelty of the mentioned approaches is that they do not require to learn target-specific appearance models and collect samples to distinguish different people from each other. The methods are evaluated on large-scale tracking datasets. State-of-the-art performance is achieved using motion metadata such as person bounding box and camera number, and shows better associations for the challenging exit-entry cases.     

Visual tracking via dynamic tensor analysis with mean update

The appearance model is an important issue in the visual tracking community. Most subspace-based appearance models focus on the time correlation between the image observations of the object, but the spatial layout information of the object is ignored. This paper proposes a robust appearance model for visual tracking which effectively combines the spatial and temporal eigen-spaces of the object in a tensor reconstruction way. In order to capture the variations in object appearance, an incremental updating strategy is developed to both update the eigen-space and mean of the object. Experimental results demonstrate that, compared with the state-of-the-art appearance models in the tracking literature, the proposed appearance model is more robust and effective.     

Tracking in low frame rate video: a cascade particle filter with discriminative observers of different life spans

Tracking object in low frame rate video or with abrupt motion poses two main difficulties which most conventional tracking methods can hardly handle: 1) poor motion continuity and increased search space; 2) fast appearance variation of target and more background clutter due to increased search space. In this paper, we address the problem from a view which integrates conventional tracking and detection, and present a temporal probabilistic combination of discriminative observers of different lifespans. Each observer is learned from different ranges of samples, with different subsets of features, to achieve varying level of discriminative power at varying cost. An efficient fusion and temporal inference is then done by a cascade particle filter which consists of multiple stages of importance sampling. Experiments show significantly improved accuracy of the proposed approach in comparison with existing tracking methods, under the condition of low frame rate data and abrupt motion of both target and camera.     

视觉目标跟踪十年研究进展

视觉目标跟踪指在一个视频序列中,给定第一帧目标区域,在后续帧中自动匹配到该目标区域的任务.通常来说,由于场景遮挡、光照变化、物体本身形变等复杂因素,目标与场景的表观会发生剧烈的变化,这使得跟踪任务本身面临极大的挑战.在过去的十年中,随着深度学习在计算机视觉领域的广泛应用,目标跟踪领域也迅速发展,研究人员提出了一系列优秀...     

An online spatio-temporal tensor learning model for visual tracking and its applications to facial expression recognition

Robust visual tracking remains a technical challenge in real-world applications, as an object may involve many appearance variations. In existing tracking frameworks, objects in an image are often represented as vector observations, which discounts the 2-D intrinsic structure of the image. By considering an image in its actual form as a matrix, we construct the 3rd order tensor based object representation to preserve the spatial correlation within the 2-D image and fully exploit the useful temporal information. We perform incremental update of the object template using the N-mode SVD to model the appearance variations, which reduces the influence of template drifting and object occlusions. The proposed scheme efficiently learns a low-dimensional tensor representation through adaptively updating the eigenbasis of the tensor. Tensor based Bayesian inference in the particle filter framework is then utilized to realize tracking. We present the validation of the proposed tracking system by conducting the real-time facial expression recognition with video data and a live camera. Experiment evaluation on challenging benchmark image sequences undergoing appearance variations demonstrates the significance and effectiveness of the proposed algorithm.     

Object tracking via collaborative multi-task learning and appearance model updating

In this paper, we propose a novel visual tracking algorithm using the collaboration of generative and discriminative trackers under the particle filter framework. Each particle denotes a single task, and we encode all the tasks simultaneously in a structured multi-task learning manner. Then, we implement generative and discriminative trackers, respectively. The discriminative tracker considers the overall information of object to represent the object appearance; while the generative tracker takes the local information of object into account for handling partial occlusions. Therefore, two models are complementary during the tracking. Furthermore, we design an effective dictionary updating mechanism. The dictionary is composed of fixed and variational parts. The variational parts are progressively updated using Metropolis–Hastings strategy. Experiments on different challenging video sequences demonstrate that the proposed tracker performs favorably against several state-of-the-art trackers.     

协同结构稀疏重构的判别性视觉跟踪

基于稀疏表示的表观似然模型在目标跟踪领域具有广泛的应用,但是这种单一产生式目标表观模型并未考虑完整的判别性结构信息,容易受复杂背景的干扰。为了缓解由该问题造成的目标跟踪漂移,提出了一种目标表观字典和背景字典协同结构稀疏重构优化的视觉跟踪方法。通过构建一个有判别力的基于稀疏表示的表观似然模型,实现了对目标表观模型更为准确的描述。通过合理选择约束候选目标区域和候选背景区域的稀疏系数,在表观似然模型中引入判别式信息,以进一步揭示候选目标区域的潜在相关性和候选背景区域的结构关系,从而更加准确地学习候选目标区域的表观模型。大量有挑战性的视频序列上的实验结果验证了算法在复杂背景下跟踪的鲁棒性,与其他相关算法的对比实验也体现了该算法的优越性。     

Online adaptive radial basis function networks for robust object tracking

Visual tracking has been a challenging problem in computer vision over the decades. The applications of visual tracking are far-reaching, ranging from surveillance and monitoring to smart rooms. In this paper, we present a novel online adaptive object tracker based on fast learning radial basis function (RBF) networks. Pixel based color features are used for developing the target/object model. Here, two separate RBF networks are used, one of which is trained to maximize the classification accuracy of object pixels, while the other is trained for non-object pixels. The target is modeled using the posterior probability of object and non-object classes. Object localization is achieved by iteratively seeking the mode of the posterior probability of the pixels in each of the subsequent frames. An adaptive learning procedure is presented to update the object model in order to tackle object appearance and illumination changes. The superior performance of the proposed tracker is illustrated with many complex video sequences, as compared against the popular color-based mean-shift tracker. The proposed tracker is suitable for real-time object tracking due to its low computational complexity.     

深度学习的目标跟踪算法综述

目标跟踪是利用一个视频或图像序列的上下文信息,对目标的外观和运动信息进行建模,从而对目标运动状态进行预测并标定目标位置的一种技术,是计算机视觉的一个重要基础问题,具有重要的理论研究意义和应用价值,在智能视频监控系统、智能人机交互、智能交通和视觉导航系统等方面具有广泛应用。大数据时代的到来及深度学习方法的出现,为目标跟踪的研究提供了新的契机。本文首先阐述了目标跟踪的基本研究框架,从观测模型的角度对现有目标跟踪的历史进行回顾,指出深度学习为获得更为鲁棒的观测模型提供了可能;进而从深度判别模型、深度生成式模型等方面介绍了适用于目标跟踪的深度学习方法;从网络结构、功能划分和网络训练等几个角度对目前的深度目标跟踪方法进行分类并深入地阐述和分析了当前的深度目标跟踪方法;然后,补充介绍了其他一些深度目标跟踪方法,包括基于分类与回归融合的深度目标跟踪方法、基于强化学习的深度目标跟踪方法、基于集成学习的深度目标跟踪方法和基于元学习的深度目标跟踪方法等;之后,介绍了目前主要的适用于深度目标跟踪的数据库及其评测方法;接下来从移动端跟踪系统,基于检测与跟踪的系统等方面深入分析与总结了目标跟踪中的最新具体应用情况,最后对深度学习方法在目标跟踪中存在的训练数据不足、实时跟踪和长程跟踪等问题进行分析,并对未来的发展方向进行了展望。