UCF YouTube Research Articles

Human Activity Classification Using the 3DCNN Architecture

Interest in utilizing neural networks in a variety of scientific and academic studies and in industrial applications is increasing. In addition to the growing interest in neural networks, there is also a rising interest in video classification. Object detection from an image is used as a tool for various applications and is the basis for video classification. Identifying objects in videos is more difficult than for single images, as the information in videos has a time continuity constraint. Common neural networks such as ConvLSTM (Convolutional Long Short-Term Memory) and 3DCNN (3D Convolutional Neural Network), as well as many others, have been used to detect objects from video. Here, we propose a 3DCNN for the detection of human activity from video data. The experimental results show that the optimized proposed 3DCNN provides better results than neural network architectures for motion, static and hybrid features. The proposed 3DCNN obtains the highest recognition precision of the methods considered, 87.4%. In contrast, the neural network architectures for motion, static and hybrid features achieve precisions of 65.4%, 63.1% and 71.2%, respectively. We also compare results with previous research. Previous 3DCNN architecture on database UCF Youtube Action worked worse than the architecture we proposed in this article, where the achieved result was 29%. The experimental results on the UCF YouTube Action dataset demonstrate the effectiveness of the proposed 3DCNN for recognition of human activity. For a more complex comparison of the proposed neural network, the modified UCF101 dataset, full UCF50 dataset and full UCF101 dataset were compared. An overall precision of 82.7% using modified UCF101 dataset was obtained. On the other hand, the precision using full UCF50 dataset and full UCF101 dataset was 80.6% and 78.5%, respectively.

A Hybrid Duo-Deep Learning and Best Features Based Framework for燗ction燫ecognition

Human Action Recognition (HAR) is a current research topic in the field of computer vision that is based on an important application known as video surveillance. Researchers in computer vision have introduced various intelligent methods based on deep learning and machine learning, but they still face many challenges such as similarity in various actions and redundant features. We proposed a framework for accurate human action recognition (HAR) based on deep learning and an improved features optimization algorithm in this paper. From deep learning feature extraction to feature classification, the proposed framework includes several critical steps. Before training fine-tuned deep learning models – MobileNet-V2 and Darknet53 – the original video frames are normalized. For feature extraction, pre-trained deep models are used, which are fused using the canonical correlation approach. Following that, an improved particle swarm optimization (IPSO)-based algorithm is used to select the best features. Following that, the selected features were used to classify actions using various classifiers. The experimental process was performed on six publicly available datasets such as KTH, UT-Interaction, UCF Sports, Hollywood, IXMAS, and UCF YouTube, which attained an accuracy of 98.3%, 98.9%, 99.8%, 99.6%, 98.6%, and 100%, respectively. In comparison with existing techniques, it is observed that the proposed framework achieved improved accuracy.

A framework of human action recognition using length control features fusion and weighted entropy-variances based feature selection

In this article, we implement an action recognition technique based on features fusion and best feature selection. In the proposed method, HSI color transformation is performed in the first step to improve the contrast of video frames and then extract their motion features by optical flow algorithm. The frames fusion approach extracts the moving regions that find out by optical flow. After that, extract shape and texture features fused by a new parallel approach name length control features. A new Weighted Entropy-Variances approach is applied to a combined vector and selects the best of them for classification. Finally, features are passed in M-SVM for final features classification into relevant human actions. The experimental process is conducted in four famous action datasets- Weizmann, KTH, UCF Sports, and UCF YouTube, with recognition rate 97.9%, 100%, 99.3%, and 94.5%, respectively. Experimental results show that the proposed scheme performed significantly sound output concerning listed methods.

Accurate Physical Activity Recognition using Multidimensional Features and Markov Model for Smart Health Fitness

Recent developments in sensor technologies enable physical activity recognition (PAR) as an essential tool for smart health monitoring and for fitness exercises. For efficient PAR, model representation and training are significant factors contributing to the ultimate success of recognition systems because model representation and accurate detection of body parts and physical activities cannot be distinguished if the system is not well trained. This paper provides a unified framework that explores multidimensional features with the help of a fusion of body part models and quadratic discriminant analysis which uses these features for markerless human pose estimation. Multilevel features are extracted as displacement parameters to work as spatiotemporal properties. These properties represent the respective positions of the body parts with respect to time. Finally, these features are processed by a maximum entropy Markov model as a recognition engine based on transition and emission probability values. Experimental results demonstrate that the proposed model produces more accurate results compared to the state-of-the-art methods for both body part detection and for physical activity recognition. The accuracy of the proposed method for body part detection is 90.91% on a University of Central Florida’s (UCF) sports action dataset and, for activity recognition on a UCF YouTube action dataset and an IM-DailyRGBEvents dataset, accuracy is 89.09% and 88.26% respectively.

A visual cognizance based multi-resolution descriptor for human action recognition using key pose

Human activity recognition using videos sequences is a well-known phenomenon which has many real-life applications such as daily assistive living, security and surveillance, patient monitoring, robotics, and sports analysis. Recently, single or still images based action recognition is becoming very popular due to spatial cues present in an image and required less computation. Hence, a robust framework is constructed by computation of textural and spatial cues of still images at multi-resolution. A fuzzy inference model is used to select the single key image from action video sequences using maximum histogram distance between stacks of frames. To represent, these key pose images the textural traits at various orientations and scales are extracted using Gabor wavelet while shape traits are computed through a multilevel approach called Spatial Edge Distribution of Gradients (SEDGs). Finally, a hybrid model of action descriptor is developed using shape and textural evidence, which is known as Extended Multi-Resolution Features (EMRFs) model. The highest classification accuracy is achieved through SVM classifier on various human action datasets: Weizmann Action (100%), KTH (95.35%), Ballet (92.75%), and UCF YouTube (96.36%). The highest accuracy achieved on these datasets are compared with similar state-of-the-art approaches and EMRFs shows superior performance.

Real-Time Pedestrian Detection Using Convolutional Neural Networks

Pedestrian detection provides manager of a smart city with a great opportunity to manage their city effectively and automatically. Specifically, pedestrian detection technology can improve our secure environment and make our traffic more efficient. In this paper, all of our work both modification and improvement are made based on YOLO, which is a real-time Convolutional Neural Network detector. In our work, we extend YOLO’s original network structure, and also give a new definition of loss function to boost the performance for pedestrian detection, especially when the targets are small, and that is exactly what YOLO is not good at. In our experiment, the proposed model is tested on INRIA, UCF YouTube Action Data Set and Caltech Pedestrian Detection Benchmark. Experimental results indicate that after our modification and improvement, the revised YOLO network outperforms the original version and also is better than other solutions.

Modeling temporal structure of complex actions using Bag-of-Sequencelets

This paper proposes a new framework for modeling temporal structures of complex human actions. Inspired by the fact that a complex action is the temporally ordered composition of sub-actions, we develop a new model named Bag-of-Sequencelets (BoS). To construct a BoS model, a video is represented as a sequence of Primitive Actions (PAs). A PA is a representative motion pattern that constitutes actions and is learned in an unsupervised manner. Representing a video as a sequence of PAs preserves their temporal order. A sequencelet is an informative sub-sequence that describes the partial structure of actions while preserving temporal relations among PAs. In a BoS model, an action is modeled as an ensemble of sequencelets. We can use sequential pattern mining to automatically learn the sequencelet without any annotation or prior knowledge of action structure. Because the BoS model has both compositional and chronological properties, it can effectively model the structures of complex actions despite intra-class variations such as viewpoint change. Experimental results show the effectiveness of the BoS model in temporal structure modeling. Applied to the Olympic sports and UCF YouTube datasets, BoS achieves greater classification accuracy than state-of-the-art methods.

Human action recognition with graph-based multiple-instance learning

A new approach to human action recognition from realistic videos is presented in this paper. First, an affine motion model is utilized to compensate background motion for the purpose of extracting dense foreground trajectories. Then, a trajectory spectral embedding is introduced to split up foreground action into multiple spatio-temporal action parts for constructing a mid-level representation. To deal with over-segmentation, a novel density discontinuity detector is proposed for the sake of generating semantically salient action parts. Finally, to handle the ambiguity in the training set, action classification is formulated within the multiple-instance learning framework, which a spatio-temporal graph model is incorporated into. Extensive experiments show that the proposed approach achieves competitive results to state of the art on UCF Sports, Kisses/Slaps, YouTube, and Hollywood datasets.

A novel hierarchical Bag-of-Words model for compact action representation

Bag-of-Words (BoW) histogram of local space-time features is very popular for action representation due to its high compactness and robustness. However, its discriminant ability is limited since it only depends on the occurrence statistics of local features. Alternative models such as Vector of Locally Aggregated Descriptors (VLAD) and Fisher Vectors (FV) include more information by aggregating high-dimensional residual vectors, but they suffer from the problem of high dimensionality for final representation. To solve this problem, we novelly propose to compress residual vectors into low-dimensional residual histograms by the simple but efficient BoW quantization. To compensate the information loss of this quantization, we iteratively collect higher-order residual vectors to produce high-order residual histograms. Concatenating these histograms yields a hierarchical BoW (HBoW) model which is not only compact but also informative. In experiments, the performances of HBoW are evaluated on four benchmark datasets: HMDB51, Olympic Sports, UCF Youtube and Hollywood2. Experiment results show that HBoW yields much more compact action representation than VLAD and FV, without sacrificing recognition accuracy. Comparisons with state-of-the-art works confirm its superiority further.

Learning Spatio-Temporal Representations for Action Recognition: A Genetic Programming Approach.

Extracting discriminative and robust features from video sequences is the first and most critical step in human action recognition. In this paper, instead of using handcrafted features, we automatically learn spatio-temporal motion features for action recognition. This is achieved via an evolutionary method, i.e., genetic programming (GP), which evolves the motion feature descriptor on a population of primitive 3D operators (e.g., 3D-Gabor and wavelet). In this way, the scale and shift invariant features can be effectively extracted from both color and optical flow sequences. We intend to learn data adaptive descriptors for different datasets with multiple layers, which makes fully use of the knowledge to mimic the physical structure of the human visual cortex for action recognition and simultaneously reduce the GP searching space to effectively accelerate the convergence of optimal solutions. In our evolutionary architecture, the average cross-validation classification error, which is calculated by an support-vector-machine classifier on the training set, is adopted as the evaluation criterion for the GP fitness function. After the entire evolution procedure finishes, the best-so-far solution selected by GP is regarded as the (near-)optimal action descriptor obtained. The GP-evolving feature extraction method is evaluated on four popular action datasets, namely KTH, HMDB51, UCF YouTube, and Hollywood2. Experimental results show that our method significantly outperforms other types of features, either hand-designed or machine-learned.

Spatio-Temporal Analysis for Human Action Detection and Recognition in Uncontrolled Environments

Understanding semantic meaning of human actions captured in unconstrained environments has broad applications in fields ranging from patient monitoring, human-computer interaction, to surveillance systems. However, while great progresses have been achieved on automatic human action detection and recognition in videos that are captured in controlled/constrained environments, most existing approaches perform unsatisfactorily on videos with uncontrolled/unconstrained conditions (e.g., significant camera motion, background clutter, scaling, and light conditions). To address this issue, the authors propose a robust human action detection and recognition framework that works effectively on videos taken in controlled or uncontrolled environments. Specifically, the authors integrate the optical flow field and Harris3D corner detector to generate a new spatial-temporal information representation for each video sequence, from which the general Gaussian mixture model (GMM) is learned. All the mean vectors of the Gaussian components in the generated GMM model are concatenated to create the GMM supervector for video action recognition. They build a boosting classifier based on a set of sparse representation classifiers and hamming distance classifiers to improve the accuracy of action recognition. The experimental results on two broadly used public data sets, KTH and UCF YouTube Action, show that the proposed framework outperforms the other state-of-the-art approaches on both action detection and recognition.

Human action recognition using weighted pooling

Pooling strategies, such as max pooling and sum pooling, have been widely used to obtain the global representations for action videos. However, these pooling strategies have several disadvantages. First, they are easily affected by unwanted background local features, the absence of discriminative local features and the times of actions periodically performed by actors. Second, most pooling strategies only use local features to build the global representation that captures little mid-level features for action representation. In this study, the authors propose a novel weighted pooling strategy based on actionlets representation for action recognition. The actionlets are defined as the movements of large bodies such as legs, arms and head, which capture rich mid-level features for action representation. Besides, the authors’ method also incorporates the distribution information of actionlets into pooling procedure. Specifically, a pooling weight, which determines the importance of actionlet on the final video representation, is assigned to each actionlet. To learn the weight, they propose a novel discriminative learning algorithm to capture the discriminative information for pooling operation. They evaluate their weighted pooling on three datasets: KTH actions dataset, UCF sports dataset and Youtube actions dataset. Experimental results show the effectiveness of the proposed method.

Realistic action recognition via sparsely-constructed Gaussian processes

Realistic action recognition has been one of the most challenging research topics in computer vision. The existing methods are commonly based on non-probabilistic classification, predicting category labels but not providing an estimation of uncertainty. In this paper, we propose a probabilistic framework using Gaussian processes (GPs), which can tackle regression problems with explicit uncertain models, for action recognition. A major challenge for GPs when applied to large-scale realistic data is that a large covariance matrix needs to be inverted during inference. Additionally, from the manifold perspective, the intrinsic structure of the data space is only constrained by a local neighborhood and data relationships with far-distance usually can be ignored. Thus, we design our GPs covariance matrix via the proposed ℓ1 construction and a local approximation (LA) covariance weight updating method, which are demonstrated to be robust to data noise, automatically sparse and adaptive to the neighborhood. Extensive experiments on four realistic datasets, i.e., UCF YouTube, UCF Sports, Hollywood2 and HMDB51, show the competitive results of ℓ1-GPs compared with state-of-the-art methods on action recognition tasks.

Weakly-Supervised Cross-Domain Dictionary Learning for Visual Recognition

We address the visual categorization problem and present a method that utilizes weakly labeled data from other visual domains as the auxiliary source data for enhancing the original learning system. The proposed method aims to expand the intra-class diversity of original training data through the collaboration with the source data. In order to bring the original target domain data and the auxiliary source domain data into the same feature space, we introduce a weakly-supervised cross-domain dictionary learning method, which learns a reconstructive, discriminative and domain-adaptive dictionary pair and the corresponding classifier parameters without using any prior information. Such a method operates at a high level, and it can be applied to different cross-domain applications. To build up the auxiliary domain data, we manually collect images from Web pages, and select human actions of specific categories from a different dataset. The proposed method is evaluated for human action recognition, image classification and event recognition tasks on the UCF YouTube dataset, the Caltech101/256 datasets and the Kodak dataset, respectively, achieving outstanding results.

Space-time Facet Model for Human Activity Classification

This paper presents a novel space-time feature-based human activity analysis system. We detect Space Time Interest Points (STIP) and generate their description based on the facet model. The proposed approach detects interest points in video data using the three-dimensional facet model efficiently. Then we describe each interest point by three-dimensional Haar wavelet transform and time derivatives of different order obtained from said facet model. Here we represent each video clip following the bag-of-words approach by learning feature specific dictionary. Finally, classification is done using non-linear SVM with χ 2 -kernel. We evaluate the performance of our system on standard datasets like Weizmann, KTH, UCF sports, ICD, UCF YouTube, and UCF50 and get better, or at least comparable results compared to other state-of-the-art systems.

Matching mixtures of curves for human action recognition

A learning-based framework for action representation and recognition relying on the description of an action by time series of optical flow motion features is presented. In the learning step, the motion curves representing each action are clustered using Gaussian mixture modeling (GMM). In the recognition step, the optical flow curves of a probe sequence are also clustered using a GMM, then each probe sequence is projected onto the training space and the probe curves are matched to the learned curves using a non-metric similarity function based on the longest common subsequence, which is robust to noise and provides an intuitive notion of similarity between curves. Alignment between the mean curves is performed using canonical time warping. Finally, the probe sequence is categorized to the learned action with the maximum similarity using a nearest neighbor classification scheme. We also present a variant of the method where the length of the time series is reduced by dimensionality reduction in both training and test phases, in order to smooth out the outliers, which are common in these type of sequences. Experimental results on KTH, UCF Sports and UCF YouTube action databases demonstrate the effectiveness of the proposed method.

Graph-based approach for human action recognition using spatio-temporal features

Due to the exponential growth of the video data stored and uploaded in the Internet websites especially YouTube, an effective analysis of video actions has become very necessary. In this paper, we tackle the challenging problem of human action recognition in realistic video sequences. The proposed system combines the efficiency of the Bag-of-visual-Words strategy and the power of graphs for structural representation of features. It is built upon the commonly used Space–Time Interest Points (STIP) local features followed by a graph-based video representation which models the spatio-temporal relations among these features. The experiments are realized on two challenging datasets: Hollywood2 and UCF YouTube Action. The experimental results show the effectiveness of the proposed method.

Action Recognition Using Multilevel Features and Latent Structural SVM

We first propose a new low-level visual feature, called spatio-temporal context distribution feature of interest points, to describe human actions. Each action video is expressed as a set of relative XYT coordinates between pairwise interest points in a local region. We learn a global Gaussian mixture model (GMM) (referred to as a universal background model) using the relative coordinate features from all the training videos, and then we represent each video as the normalized parameters of a video-specific GMM adapted from the global GMM. In order to capture the spatio-temporal relationships at different levels, multiple GMMs are utilized to describe the context distributions of interest points over multiscale local regions. Motivated by the observation that some actions share similar motion patterns, we additionally propose a novel mid-level class correlation feature to capture the semantic correlations between different action classes. Each input action video is represented by a set of decision values obtained from the pre-learned classifiers of all the action classes, with each decision value measuring the likelihood that the input video belongs to the corresponding action class. Moreover, human actions are often associated with some specific natural environments and also exhibit high correlation with particular scene classes. It is therefore beneficial to utilize the contextual scene information for action recognition. In this paper, we build the high-level co-occurrence relationship between action classes and scene classes to discover the mutual contextual constraints between action and scene. By treating the scene class label as a latent variable, we propose to use the latent structural SVM (LSSVM) model to jointly capture the compatibility between multilevel action features (e.g., low-level visual context distribution feature and the corresponding mid-level class correlation feature) and action classes, the compatibility between multilevel scene features (i.e., SIFT feature and the corresponding class correlation feature) and scene classes, and the contextual relationship between action classes and scene classes. Extensive experiments on UCF Sports, YouTube and UCF50 datasets demonstrate the effectiveness of the proposed multilevel features and action-scene interaction based LSSVM model for human action recognition. Moreover, our method generally achieves higher recognition accuracy than other state-of-the-art methods on these datasets.

Combining appearance and structural features for human action recognition

In this paper, we propose to integrate structural information with appearance features for human action recognition. In local representations based on detected spatio-temporal interest points (STIPs), the layout of STIPs carries important cues of motion structures in video sequences, and is assumed to contain complementary information to appearance features. We aim to incorporate structures into the description of STIPs by combing with appearance features for action representation.Based on the previous work of the 3D shape context, we present an optimised version of 3D shape context to encode the layout information of STIPs. By combining the proposed optimised 3D shape context with appearance descriptors, e.g., HOG3D and 3D gradients, we provide a more informative and discriminative description of STIPs for action classification. To validate the proposed descriptor, we have conducted extensive experiments on the KTH and the UCF YouTube datasets. The results prove that the optimised 3D shape context offers complementary information to appearance features, showing its effectiveness for action representation; moreover, the proposed descriptor yields comparable results with the state-of-the-art methods.

Content-based retrieval of human actions from realistic video databases

Due to the increasing amount of video data available in various databases, on the Internet and elsewhere, new methods of managing these data are required, leading to the development of content-based video retrieval systems. We explore several recently developed action representation and information retrieval techniques in a human action retrieval system. These techniques include various means of local feature extraction; soft-assignment clustering; Bag-of-Words, vocabulary guided and spatio-temporal pyramid matches for action representation; SVMs and ABRS-SVMs for relevance feedback. Successful application of relevance feedback in particular will result in far more practical systems. We evaluate the performance of several combinations of the above techniques in three realistic action datasets: UCF Sports, UCF YouTube and HOHA2.