E-Learning and Games

Virtual reality technology shows serious potential in many fields, such as cinemtic entertainment, professional training, Healthcare and clinical therapies, etc. In this paper, we propose a novel human balance capability evaluation method, which is based on crossing bridge virtual scene and video analysis. We have sampled the crossing bridge movement video of two groups of volunteers with balance ability differences, and then we proposed a balance ability classification algorithm via barycentric shifts model statistical analysis. The small sample experiment shows that our method can accurately identify the possible candidates with balance ability abnormality.

In this paper, we introduce an integrated framework to simulate waiting halls with mass passengers. We design the framework for the special purpose of passenger safety investigation. It integrates virtual waiting hall environment, mass virtual passengers with heterogeneous behavior and motion, and also editable scenarios to conduct virtual passengers. So that different situations could easily be initialized and simulated to see how they are developed and evolved as time goes on. We also introduce a behavioral decision and execution method which is embedded in our framework. It supports both regular crowded passenger behaviors and emergency passenger behaviors. Results show that simulated passengers have realistic behavior and act heterogeneously in a crowded waiting hall environment under both normal and emergent scenarios.

Extracting geographical information from geospatial data has a high priority for all geographic activities which needs powerful tools of visualization. In this paper, a holographic approach is proposed to bridge the gap between mixed reality and virtual world rendering. It envisages the georeferenced raster-based data which are integrated into a virtual world which assembles data from different sources and could be projected them into the real world in order to enhance extracting visually the interesting geographical information. Experiment achieved on the HoloLens. This method will add holographic to improve data observation and to enrich the geographic edutainment.

In this paper, the design and development of the ARGBL multiplayer game is shown and lessons learned for each of the design and development stages, are stated with recommendations for future ARGBL multiplayer projects.

Imageless navigation systems (INS) in orthopaedics have been used to improve the outcomes of several orthopaedic procedures such as total hip replacement [1, 2]. However, the increased surgical times and the associate learning curve discourage surgeons from using navigation systems in their theatres [2]. This paper presents a Mixed Reality (MR) simulator that helps surgeons acquire the infrared based navigation skills before performing it in reality. A group of 7 hip surgeons tried the application, expressing their satisfaction with all the features and confirmed that the simulator represents a cheaper and faster option to train surgeons in the use of INS than the current learning methods.

Traditionally, users used to adopt external device as the controller for the virtual environment. Nowadays we are more interested in natural interaction, such as visual interaction within the virtual reality (VR). More natural human-machine interaction methods would be employed to improve the accessibility of interaction when the user immerses in the virtual or augmented reality. This paper introduces our work on a project called “Collision as Trigger (CAT)” in the mobile virtual reality field, cooperating with a VR helmets vender, a mobile phone manufacturer and a digital entertainment content provider. Our outcomes include stereopsis, motion sensing, scene switching based on collisions as the trigger, as well as some experience improvements on the noise suppression.

Many archaeological sites are less popular by visits amongst the younger group and overall less popular than majority of other heritage sites. They are often not enhanced by supporting medium like in museums or historic buildings. Many augmented reality (AR) systems have been developed for archaeological sites and proved to benefit user engagement. However, most result in the superimposition of virtual reconstructions of the site and very little interaction. In this paper, we demonstrate the development of a location-based treasure hunt AR app, Avebury Portal, for the heritage site; Avebury in England. Avebury Portal uses puzzles with the environment to give clues, and a narrative that responds to the user’s location. We developed Avebury Portal with Unity Engine and Vuforia to demonstrate the effectiveness of using AR to enhance visitors’ experiences on learning.

This paper explores the benefits of a sales promotion in the aviation industry known as Frequent-Flyer Program. Four well known Chinese FFPs are chosen as a benchmark to assess their gamification effect with a focus on game sophistication and costumer’s experience. A data-driven approach is employed to analyze gamification techniques in FFPs such as tiers system and points system. The results show that the degree of game sophistication is relatively low due to its non-game context. The present contribution illustrates how tiers system and points system offer fun-game and serious-game experience respectively. It also shows an advantage of its harmonious combination to attract more potential customers and retain the frequent flyer customers.

In this study, a serious game was developed to enable typically developing to learn the most appropriate method of conversation with autism. Autism tend to have difficulty with interpersonal relationships because they find it difficult to understand conversational implicatures. As a solution to this problem, we propose a serious game to enable typically developing to learn how to speak to autism in a way they can understand, including how best to convey conversational implicatures. This serious game simulates the experience of autism of being unable to understand conversational implicatures, using conversations based on real life examples, in order to teach players why autism cannot understand what they mean and how they can paraphrase their message. To verify the efficacy of this serious game, an experiment was conducted with 56 students. Experimental results suggested that the system is effective for learning how to speak especially among typically developing who were rarely involved with autism.

In recent years, with the development of computer science and graphics technology, the Gamification of training industry has become increasingly prosperous. The virtual driving training system in this paper takes the post-90s Chinese youth as the main research object, studies their social environment condition, work habits, cultural background and value identity change trend, using qualitative investigation method and quantitative experimental method, to record and analysis the users’ emotional changes. Finally, applying the theory and method of cognitive psychology and situated learning, to make a comprehensive comparison and research on the user demand model and experimental data, and obtained the guidance criteria for the system development and evaluation.

Aimed at the Chinese aging population, companion robots with entertainment and communication functions are popular and useful for the daily life of the elderly. To make robots more reliable and socially acceptable in real scenarios, a better affective interaction with intelligent emotions and adaptive behaviors of companion robots is needed. Through the listed status of human-robot interaction and learning models, the advantages and limitations of companion robots are shown distinctly. In the end, the problems that might exist in technical and ethical at present are proposed, in order to provide references for the further researches.

We present a proposal for the application of gamification strategies to an introductory course in algorithms and programming, which is aimed at different university engineering programs. The purpose of the course is that students acquire skills in the development of algorithms and their implementation in a programming language to solve problems in various domains. The activities proposed for the course rely mainly on a platform which allows the presentation and evaluation of these activities. Performance in the field is low, with a rather high failure rate. Our purpose is to propose some gamification strategies in the classroom, based on mechanisms implemented both in the platform and in a mobile application that could serve as support for the course.

Indoor scene reconstruction from point cloud data provided by Terrestrial laser scanning (TLS) has become an issue of major interest in recent years. However, the raw scanned indoor scene is always complex with severe noise, outliers and incomplete regions, which produces more difficulties for indoor scene modeling. In this paper, we presented an automatic approach to reconstruct the structure of indoor scene from point clouds acquired by registering several scans. Our method first extracts different candidate walls by separating the indoor scene into different planes based on normal variation. Then the boundary of those candidate walls are obtained by projecting them onto 2D planes. We classify the walls into exterior wall and interior wall by clustering. After distinguishing the 3D points belonging to exterior walls, a simple strategy is generated to refine the 3D model of wall structure. The methodology has been tested on three real datasets, which constitute of different varieties of indoor scenes. The results derived reveal that the indoor scene could be correctly extracted and modeled.

Accurate rendering a real-world object has been a long-standing challenge in computer graphics area. The most popular method is BRDF, it is valid for opaque materials, such as metals, but it fails for translucent materials, such as skin. In order to render human skin faster and better, we propose a layered method D-BRDF, which divides the face model into three layers: sebum, epidermis and dermis, and combines the ambient light, specular reflection of sebum, diffuse reflection of the epidermis and subsurface scattering of the dermis to get the sum of light intensity and the final rendering effect. We experiment on several models, and the results show that it is more effective and faster than BRDF working on translucent models. The effects are downy and transparent, especially in the ear, cheek and other details. And it can be widely used in other related areas.

Nowdays, VoD (video-on-demand) has become a wide-used technology in m-learning. In m-learning VoD systems, we need to allocate appropriate bandwidth for streaming media servers with the aim of optimizing the user experience and reducing the service cost. In this paper, a queue-based bandwidth allocation method for streaming media servers in m-learning VoD system is proposed. Firstly, it analyzes the user historical learning logs to mine the user behavior characteristics. Secondly, it utilizes the queueing theory to establish a bandwidth resource allocation model for streaming media servers. Thirdly, it predicts the user arrival rate in real-time, allocates appropriate bandwidth resource dynamically by the bandwidth resource allocation model, so as to solve the bandwidth resource allocation irrationality problem. Finally, the simulation results have proved the correctness and effectiveness of the proposed bandwidth resource allocation method, which can improve the bandwidth resource utilization and reduce the service rejection rate.

A point cloud hole repairing method based on edge-preserving projection is proposed in order to maintain sharp features of holes. First, the hole boundary points are acquired by quadrants and angles, then connections among 3D hole boundary points are projected onto the two-dimensional plane using the edge-preserving projection. Secondly, the two-dimensional region obtained from projection is subjected to point collectivization for obtaining filling points, at the same time, the radial-basis interpolation mapping technique is used to construct the hole surface according to the 3D hole boundary point. Finally, the two-dimensional filling points are reflected onto the surface of the constructed hole to complete hole repairing. Experimental results show that our method can effectively repair the hole of the point cloud and restore the sharp features of the hole.

The repairing of 3D point cloud holes has an important meaning to ensure the integrity of cloud data. We present a slice-based repairing method for 3D point cloud in this paper. Firstly, the model is horizontal sliced and each slice is projected on a two-dimensional plane, then the band-shaped points obtained during projection are clustered to select boundary points of the hole. Combined with optimal fitting points, the hole repairing point sets in the projection layer are re-sampled based on the cubic B-spline curve to fit boundary points. Finally, all hole repairing point sets in the projection layer are combined in 3D space to finish the entire hole recovery. The experimental results show that the proposed method can effectively repair complex holes for various point cloud models.

Total variation denoising model is vulnerable to the influence of the gradient and often loses the image details. Aiming at this shortcoming, an improved total variation denoising model is proposed to recover the damaged additive Gaussian noise image. First, guided filtering and impulse filtering are used to preprocess noisy images; second, the adaptive norm parameter is selected by the edge detection operator; third, the horizontal and vertical weight values are selected by adaptive method; Finally, the image processed by non-local means filter replaces the noisy image to modify the fidelity term in the method. Experiments show that the improved total variation denoising model can remove the noise and can keep the texture and edge of the image better as well.

Multispectral image encoding/decoding methods using spectral dictionary learning and sparse representation to fully exploit spectral features are proposed. In the scheme, K-SVD is first adopted for training a redundant dictionary from typical similar spectra. Then the sparse representative coefficients of each spectrum are obtained by the dictionary for spectral redundancy removal. Finally the equivalent nonzero sparse coefficients are quantified and stored. Experimental results show the superior spectral reconstructed performance compared with sample principal component analysis (PCA) and classical adaptive PCA at the same or even lower bit rates. Besides, the spectral dictionary learning can also be combined with compressed sensing or spatial decorrelation technologies to further expand its application.

An intrinsic co-decomposition model is presented for stereoscopic images. To build the correlation of inter-image or intra-image, the sparse subspace clustering in superpixel level and K-mean clustering in pixel level are implemented. With the constraints on correlation, stereoscopic images are decomposed simultaneously and the reflectance components with more details and higher contrasts are obtained for the edge-preserving of superpixel and the local reflectance correlation of pixels. Experiments show that the reflectance components of co-decomposition are clearer visually. Furthermore, information entropy and standard deviation of reflectance components of co-decomposition are calculated to validate the effectiveness quantitatively of the co-decomposition.

In this paper, improved scattering parameters of polarimetric synthetic aperture radar (POLSAR) image based on spatial information and Bayes rule is proposed. The spatial information of scattering parameters is introduced by using an adaptive weight window. Bayes rule is used to improve the performance of the scattering parameters. Experiments on real AIRSAR L-band fully POLSAR data are carried out, and the efficacy of the improved scattering parameters is verified.

Traditional spectral clustering (SC) employed k-means to find the cluster centers, which leads to the problem of sensitive to initialization and easily falls into local optimum. To address this issue, a novel superpixel-based immune clonal spectral clustering (ICSC) method in the spatial-polarimetric domain is proposed for PolSAR data classification. Firstly, the proposed method divides PolSAR image into superpixels, which not only considers the region homogeneity but also reduces the computational complexity. After that, combined manifold distance measures in the spatial-polarimetric domain are used to construct the similarity matrix. Finally, immune clonal algorithm (ICA) is substituted for k-means to obtain global optimum solution with large probability. Experiments results show the feasibility and efficiency of the proposed method.

Due to the serious situation of the public security in China, it has important practical meaning to make contingency plans for emergencies in advance. During the emergency evacuation, crowd will be in a non-rational state. The characteristics of the crowd aggregation are different from the usual crowd movements and easily lead to emotional contagion. However, the research on emotional contagion in crowd animation is still rare in the existing studies. From the perspective of psychology, this paper discusses the related concepts, parameters, and methods of measurement for emotional contagion, and summarizes the work of using computer vision technology in obtaining the crowd parameters. Based on these, a model of emotional contagion has been proposed. The experimental results show that the proposed model can well represent the crowd evacuation behavior and can be a new method for crowd emergency management.

Semantic human body reshaping builds a 3D body according to several anthropometric measurements, playing important roles in virtual fitting and human body design. We propose a novel part-based semantic body model for 3D body reshaping. We adopt 20 types of measurements in regard of length and girth information of body shape. Our approach takes any number (1–20) of measurements as input, and generates a 3D human body. Firstly, all missing measurements are estimated from known measurements using a correlation-based method. Then, based on our proposed semantic model, we learn corresponding semantic body parameters which determine a 3D body from measurements. Our model is trained using a database of 4000 registered body meshes which are fitted with scans of real human bodies. Through experiments, we compare our approach with previous methods and show the advantages of our model.

In recent years, there has been an important growth of online gaming. Today’s Massively Multiplayer Online Games (MMOGs) can contain millions of synchronous players scattered across the world and participating with each other within a single shared game. Traditional Client/Server architectures of MMOGs exhibit different problems in scalability, reliability, and latency, as well as the cost of adding new servers when demand is too high. P2P architecture provides considerable support for scalability of MMOGs. It also achieves good response times by supporting direct connections between players. In this paper, we have proposed a novel dynamic load balancing for massively multiplayer online games (MMOGs) based this hybrid Peer-to-Peer architecture. We have divided the game world space into several regions. Each region in the game world space is controlled and managed by using both a super-peer and a clone-super-peer. The region’s super-peer is responsible for distributing the game update among the players inside the region, as well as managing the game communications between the players. However, the clone-super-peer is responsible for controlling the players’ migration from one region to another, in addition to be the super-peer of the region when the super-peer leaves the game. We have designed and evaluated the dynamic load balancing for MMOGs based on hybrid P2P architecture. We have used OPNET Modeler 18.0 to simulate and evaluate the proposed system. Our dynamic load balancer is responsible for distributing the load among the regions in the game world space. The position of the load balancer is located between the game server and the regions. The results, following extensive experiments, show that low delay and higher traffic communication can be achieved using dynamic load balancing for MMOGs based on hybrid P2P system.

The structure information of indoor scene is necessary for a robot who works in a room. In order to achieve structure of an indoor scene, a slice-guided method of indoor scene structure retrieving is proposed in this paper. We present a slicing based approach that transforms three-dimensional (3D) segmentations into two-dimensional (2D) segmentation and segments different kinds of primitive shapes while keeping the global topology structure of the indoor scene. The global topology structure is represented by a graph. The graph is compared with the given indoor scene template. The matched objects and the topology relation between them are finally presented. Our experiment results show that the proposed method performs well on several typical indoor scenes, even if some data are missing.

In this paper, we introduce a deep reinforcement learning approach for autonomous car racing based on the Deep Deterministic Policy Gradient (DDPG). We start by implementing the approach of DDPG, and then experimenting with various possible alterations to improve performance. In particular, we exploit two strategies: the action punishment and multiple exploration, to optimize actions in the car racing environment. We evaluate the performance of our approach on the Car Racing dataset, the experimental results demonstrate the effectiveness of the proposed approach.

In this paper, an evolutionary algorithm based on bi-goal is proposed for many-objective optimization. We first provide a new proximity estimation, ensuring the convergence of algorithm. Afterwards, a new sharing function with a novel discriminator is employed to improve the diversity. The dominance-based environmental selection is applied in bi-goal space, which is expected to archive a good balance between convergence and diversity. The experimental results show that the proposed method can work well on most instances considered in this study, demonstrating that it is very competitive for solving many-objective optimization problems.

Motion retrieval has important practical value for the reuse of motion capture data. However, it is a challenging task to represent the motion data effectively due to the complexity of the motion data structure. As graph models is an effective way to represent structured data. This paper proposes a new method for human motion retrieval based on graph model. First, a method of graph model constructing based on Maximum Range of the Distance (MRD) is proposed. The MRD is used to select the joint pairs that are deemed important for a given motion, and different motions have different graph model structures. After that, similar motions can be retrieved by matching the similarity of the attributes of graph model. In the process of motion retrieval, cosine similarity is defined to measure the similarity of graph models. The experimental results show that the method proposed in this article is better than the previous methods of motion retrieval in many ways.

The rise of skeletal skinning technology has provided great convenience for animators. At the same time, it improves the efficiency of animation production. However, the deformation resulting from this technology suffers from some undesirable effects, which require manual improvement. In this paper, we propose an approach addressing the problem of creating believable mesh-based skin deformation. In this approach, the skin is first deformed with a classic linear blend skinning approach, which usually lead to artifacts like the candy-wrapper effect or volume loss. Then we enforce the geometric constraints which displace the positions of the vertices to mimic the behavior of the skin and achieve effects like volume preservation. At last, we adopt the finite element method to handle large deformed elements which could accelerate the system’s convergence rate. This approach is easy to implement and has a high skinning efficiency without affecting the simulating effect.

There are a large number of multi-objective optimization problems in real-world applications, like in games, that need to be solved in real time. In order to meet this pressing need, we suggests a method of parallelizing the multi-objective evolutionary algorithm based on decomposition (MOEA/D). Furthermore, a novel task decomposition strategy and scalarizing method without the ideal point are proposed for meeting the requirements of real-time and precision of the game. By combining the novel scalarizing function and GPU-based CUDA technology with the MOEA/D, a parallel MOEA/D for real-time multi-objective optimization problems is developed, namely P-MOEA/D. Experimental studies on ZDT and DTLZ benchmark problems suggest that the P-MOEA/D algorithm is efficient and fast.

According to the characteristics of underwater target tracking, extended Kalman filter (EKF) algorithm was applied to underwater bearing-only and bearing-Doppler non-maneuverable target tracking problem. EKF is recursive Bayesian filter algorithm based on the linearization of the nonlinearities in the target state and the measurement equations. To ensure the observability in passive target tracking, we use single maneuvering observer. The simulation results show the suitability and effectiveness of the EKF algorithm to the single non-maneuverable target.

In this paper, a motion-driven system was designed to combine the motion of human performers with physical simulations in order to generate aesthetic visual effects that respond to the performers in realtime. The system consists of two major components: the motion data acquisition and the visual effects feedback. We implement the motion data acquisition module based on infrared sensors which provides realtime performance with both 2D and 3D outputs. The visual effects feedback module is designed in charge of producing aesthetic effects based on the realtime motion data. We evaluated the effectiveness of our framework on several performing art shows. The results suggest that our system is capable of enhancing the traditional electronic art effects.

The heterogenous areas of the polarimetric synthetic aperture radar (PolSAR) image are hardly to be classified into semantic homogenous regions due to the complex terrain structures. In order to overcome these disadvantages, a PolSAR image classification method is proposed based on the multi-feature learning and the topic model. The proposed method makes use of three kinds of features to formulate the visual codewords. Then, the higher level features are learned by the topic model for classification. Experimental results illustrate that the proposed method can obtain better performance than the state-of-art methods especially for the heterogenous areas.

In this paper, we present TLogic, a tangible programming tool for children aged 6–8. The tool contains two parts: tangible programming blocks and visual tasks with different difficulty levels. Children could use tangible programming blocks to complete visual tasks shown on computer screen. To evaluate our tool, a user study was conducted with 10 children. Results of user study show that the tool could reduce children’s cognitive load while solving the tasks and children are more interested in challenging tasks than easy tasks in our tool.

In response to the “Mass Entrepreneurship and Innovation”, University of Jinan has built full-covered and hierarchical entrepreneurship and innovation education. In this paper, as a case of Computer Science and Technology of University of Jinan, the school-enterprise cooperative innovation and entrepreneurship courses and case library have been proposed under the support Entrepreneurship School. The course system considers both liberal and professional knowledge, improves interest in student participation, and reinforces school-government-enterprise cooperation. Some measures such as flipped classroom and achievement quality track have been taken to contribute innovation and entrepreneurship.

The construction of the world-class university and the world-class discipline makes a clear request to improve the system and ability of internal governance in higher education institutions. However, it is a common problems in colleges and universities in China, which the lag of institutional system, the serious imbalance between administrative power and academic power, the low efficiency of management. The negative list is an internationally accepted mode of foreign investment access management, which with the characteristics of law-based authority and responsibility, bottom-line management and shared governance. On the basis of clarifying the connotation and characteristics of the negative list, this paper gives some ideas for the introduction of negative list management in reform of internal governance in colleges and universities.

This paper treats Japanese history education as one of the activities of the center called ICER (Innovation Center for Educational Resources), Kyushu University. In this activity, one of the key challenges is the development of educational materials that attract students to Japanese history. As a first step, a database will be built that contains the Japanese history knowledge for the educational material development. So, the authors have already been building a database based on Linked Data. The other research agenda is to provide e-learning materials themselves using the database that attract students to Japanese history. The use of recent ICT (Information & Communication Technology) like 3D graphics enables e-learning materials to attract the students. In this paper, the authors propose interactive web 3D contents development framework based on Linked Data for Japanese history education.

We developed a game with a purpose (GWAP), which collects structured data corresponding to adjectives to build a visual effect dictionary. In this system, new semantic links can be acquired. Under the guise of a fighting game, the system encourages the user to vote on the commonsense knowledge associated with an object, because our previous research indicated that the rules of showing appropriate visual effect according to the adjective is related to commonsense knowledge of the target object. This system displays visual effects on the target object and the data structure is updated based on user’s vote. This structured data underlies a new type of communication support system that continuously improves visual effects that modify adjectives and objects. In this paper, we discuss the structure of the visual effect dictionary through an experiment. Findings show that GWAP effectively improves the relationship between commonsense knowledge and objects, while creating new linkages via deduction.

Gamification and E-learning are the application of game-based elements and game design techniques. Many entertainment and learning platforms have applied gamification to increase motivation and engagement. Game refinement theory is a new game theory which concerns about the entertaining aspects of games using a game sophistication measure that is derived from a game progress model; it can judge the game quality. This paper analyzes the fighting game, which is a kind of video games where two on-screen characters fight with each other. The game refinement measure is employed for the assessment of the game sophistication of fighting games in different types. The analyzed results show the evolutionary changes of the fighting games. Also, it can show the experience of suitable E-learning rhythm. In the future, the human can use this method to design the target e-learning platform become more comfortable and reasonable.

Mobile learning have a capacity to respond to educational needs of learners dispersed across vast regions and cultures. Among a wide variety of popular mobile platforms. Android has become one of the optimal mobile operating systems. In this paper, we investigate the challenges of the mobile learning and propose a framework of mobile learning system. When the student log in for identification verification, they can select the recorded coursewares, join the real time class in various interactive learning methods including text, video and audio. A system with a user-friendly interface is designed, based on which experiments are performed to evaluate the effectiveness of the proposed system which can support 4 or 6 interactions at the same time.

In recent years, with the rapid development of network communication technology and the widespread use of mobile devices. Mobile learning is getting more and more attention and research by scholars, becoming a new type of learning method. Students group is the highest popularization rate of smart phones, while college students also have strong ability to learn and accept new things quickly and so on. Therefore, under the new situation, colleges and universities should further study mobile learning and construct a mobile learning system that meets the needs of young students, which can provide a useful supplement for the current student education. The article introduces the research background and the status of mobile learning research. Then the article analyzes the characteristics of mobile learning and proposes a proposal for building a mobile learning system in colleges.

There are numerous and complicated relationships among personal emotional, physiological reaction and cognitive evaluation. People can produce some negative emotions at any time and place. Especially, college students have a lot of stress such as study, interpersonal, competition, employment, so that negative emotional experience affects their common life. This article aims at the design of the interactive game on the intelligent terminal, pointing at the negative emotion of the college students. Through the combination of practice and theory at home and abroad to explore the role of interactive gameplay in emotional regulation. To provide users with a virtual platform outside real life, name it “Mood Robert.” The emotional appeal is realized by adjusting the contradiction between “ego” and “super-ego.” Sharing their emotions based on social network information communication to improve the ability to adjust themselves and reduce the impact of negative emotions on daily life.

The job-hunting and graduate school admission of college students are important tasks in universities. To investigate the impact of students’ academic achievement to their graduation whereabouts, Self-Organizing Maps is introduced in this study. Through the analysis of experiment results, the features of academic performance in different students’ graduation whereabouts segments are proposed. The findings could help educators better understand the relationship between academic performance and graduation whereabouts.

This research shows a new methodology to teach math fractions. It is based on a neuro-pedagogical approach where students have to create their own educational real-world game called “Hands on work” using recycled material. The experiments were done in two schools in Apurimac-Peru with 2 teachers and 36 students divided in two groups of 18 students. For this research we used two types of methodologies: Nominal Group for teacher’s opinion and Direct Observation for student. The results show us that, according teacher’s opinion the second stage called “Hands on Work” was the main activity to reinforce the student’s knowledge. In the other hand, 17 students who use this methodology (real world game +software game) could get 70 points between 2.3 and 4.3 min, compared with the other style (only software game) in which 13 students get 70 points between 3 and 5.5 min.

In recent years, the number of children with autism has risen sharply. It brings tremendous pain to their family. However, in many developing countries, for those autistic children, the number of rehabilitation agencies is rare, the training method is monotonous, and the rehabilitation cost is very high. Therefore, it has important practical significance to find a new effective and costless way for them. On the basis of summarizing the existing related work, this article develops a serious game prototype to assists social skills training for children with autism. The experimental results show that the emotional-based somatosensory game can significantly improve children’s interests in learning. The serious game could be a new technical solution for the adjuvant treatment for children with autism.

WebRTC (Web Real-Time-Communications) enable blended learning in universities and support the “virtual universities”. A primary challenges of the traditional technologies such as IPv4/IPv6 and TCP is the reliability. Besides, there is no free, high-quality and complete solution available that enables communication by the Web browser. In this paper, we investigate the problem of e-Learning system based on WebRTC, and propose the web-based Skyclass system on Kurento Media Server. Users may select the recorded courseware, join the real-time classroom in interactive learning methods by text, video and audio. A system with a user-friendly interface is designed, based on which experiments are performed to evaluated the effectiveness of the proposed system which can support up to 9 interactions at the same time.

Usually, the reason that virtual plant growing games cannot attract players’ attention is they cannot bring real game experience directly to players. In order to solve this problem, a plant growing game that combines virtuality with reality is designed and developed in this paper. Based on embedded hardware Arduino, Raspberry Pi and mobile terminals, game can change players’ online operating instructions into real-world action, greatly improving the authenticity and fun of game. Experiments show that the game can run effectively.

Depth cameras support working in a dark environment, and provide depth information from objects to cameras, hence have advantages over color cameras. So in this paper we adopt depth cameras to collect accurate gesture information for 3D modeling, in order to obtain accurate gesture recognition. On the depth map, we present methods of hand joint segmentation with random forest pixel classification and of gesture recognition with template matching, which provides accurate judgment for static gestures. Rotation may occur while the acquisition of hand data, so we conduct rotation correction by using SVD decomposition. Experimental results illustrate that this method provides more accurate joint segmentation, which is robust to hand rotation and achieves a recognition rate of 94.8% on ASL dataset.

In real world, there are many areas with Large images but only Small labelled (we called LS area), in there supervised and unsupervised algorithm can’t work well, but semi-supervised technology exploiting patterns both in labelled and unlabeled data to get labels can work well. The classification accuracy directly depends on the features extracted from the images. Recently, with the emergence and successful deployment of deep learning techniques for image classification, more research on getting features is directed to deep learning techniques. This paper proposes a combined semi-supervised classifier and pre-trained deep CNN model algorithm—CDLSSC (Combined Deep Learning and Semi-Supervised Classification) for LS area. The transfer learning that has been tested and verified in some areas is used to extract features in this algorithm. The method CDLSSC is evaluated on three image datasets and achieves superior performance. We apply it to the Terra-Cotta Warriors image classification area and get super results, which means that it can be used in cultural relic’s area successfully.

A novel feature-based pose estimation method for 3D faces is described in this paper. Depending on the salient crest lines which describe the rough sketch of prominent convex regions on a 3D face, the nose tip and the nose bridge are determined and used to estimate face pose without manual initialization, modeling, and training. The experimental results demonstrate that the proposed method can provide accurate, continuous and autonomous pose estimation of six degrees of freedom (DOF) for 3D faces with large pose rotation and self-occlusion.

The direct control of humanoid robot by human motion is an important aspect of current research. Most of these methods are based on additional equipments, such as Kinect, which are usually not equipped on robot. In order to avoid using these external equipments, we explored a robot controlling method only using the low-resolution camera on robot. Firstly, a stacked hourglass network is employed to obtain the accurate 2D heatmap containing positions of human joints from RGB image captured by camera on robot. Then, 3D human poses including coordinates of human body joints are estimated from 2D heatmaps by a method aiming to reconstruct 3D human poses from 2D poses. Finally, the rotation angles of robot are computed according to these 3D coordinates and are transmitted to the robot to reconstruct the original human pose. Using the NAO robot as an example, the experimental results show that the humanoid robot can imitate motions of different human actors in different scenes well while applying our method.

Human pose estimation in image is an important branch of computer vision and graphics research. In this paper, an improved modular convolution neural network is proposed to solve the problem of human pose estimation in static 2D images. A cascaded three-stage full convolutional network (FCN) can learn the non-linear mapping from image feature space to human pose space in an end-to-end way. In order to improve the accuracy of predicting joints, the method of multi-feature source fusion is adopted to improve the estimation process of the human body posture. The first two stages of the network focus on learning local image features and joints neighborhood pixel features, and these features are merged in the third stage of the network. Finally, the coordinates of human joints are obtained by regression of the merged features. In our experiments, using the strict PCP criteria on the full body pose dataset LSP, the average prediction accuracy of our method is 79.3%. In addition, using the PCKh standard on the upper body pose dataset FLIC, our method achieves an average prediction accuracy of 93% without additional training.

Cheating in examinations of the online distance education is a serious problem which may damage the fairness of exam and further undermine the credibility and reputation of certificates. In order to detect the “Ghost Writer” cheating strategy that existed in both online and offline exams, we propose the Student Identification by Face Recognition (SIFR) framework, a three layers architecture based on face recognition technique and micro-service principle, to detect the ghostwriter who takes the exam for others. In addition, we implement a prototype system based on open source projects and public cloud services. To evaluate the system, an experimental test was conducted with public data. The results indicated that the SIFR framework is feasible and the accuracy of detection is directly affected by the performance of face recognition service, which can be upgraded or replaced with better facial feature extraction module.

In this paper, a stationary directionlet (SD) domain probabilistic graphical model (PGM) for texture image segmentation is proposed. Hidden markov chain (HMC) is a good tool to capture the persistence and clustering properties of the coefficients of SD transform. The homogeneous property of texture image is described by markov random field (MRF). Combining HMC and MRF in SD domain result in SDPGM. Image segmentation based on SDPGM, which is denoted as SDPGMseg, involves inferring the maximum a posterior (MAP) solution to class labels on the coefficients of SD transform. The segmentation result can be obtained by minimizing an energy function. Experiment results show that SDPGMseg can obtain better performance especially in homogeneous regions and boundaries of different regions.

In order to improve the accuracy of hand pose estimation from a depth image, a method based on convolutional neural network (CNN) is proposed in this paper. First of all, we modify the structure of traditional CNN to recognize the 3D joint locations from a depth image. By appending some shortcuts between layers, the proposed network increases the correlation between the front and back layers. This structure can avoid the information loss caused by the simple layer-by-layer transmission, and can improve the estimation accuracy effectively. Afterwards, the estimated joint locations continue to be inputted into a support vector regression (SVR) phase. The use of SVR can introduce the constraint of local joint information, which can get rid of those abnormal estimations further. Extensive experiments show that our method enables significant performance improvement over the-state-of-arts in the accuracy of hand pose estimation.