Crowd counting is a challenging task in computer vison field and haven’t been well addressed unti... more Crowd counting is a challenging task in computer vison field and haven’t been well addressed until now. In this paper, we intend to develop an end to end multi-scale deep convolutional neural network(CNN) model that can accurately estimate the crowd count from an individual image with arbitrary crowd density and perspective. The proposed model extract multi-scale deep CNN features from the input image and regress the crwod count directly, without any post-processing . Hence our model could handle muti-scale targets well in various crowd scene. We evaluate our model on several benchmark datasets and the performance outperforms some state-of-the-art methods. What’s more, due to the end-to-end characteristics, our model demonstrates good practical application performance.
In this 'work in progress' paper, we describe research into e-learning and virtual realit... more In this 'work in progress' paper, we describe research into e-learning and virtual reality at Shanghai Jiaotong University (SJTU) and Essex University respectively. We use our experience to propose a novel virtualised learning model and architecture to support online learners and teachers to visualize important learning related information that has hitherto been invisible to users of online learning systems. The purpose of this work is to enhance the learners' experience (eg by introducing dynamic learning communities), and to help teachers by providing more efficient feedback tools to improve their teaching (eg by creating the notion of a 'virtual learner' to represent a group of similar ability learners). In undertaking this research we address challenges, such as learner community composition, virtual classroom construction, avatar modelling, collaborative Q&A machine interaction, etc and propose some novel solutions that will form our future research and expe...
Among learning based hashing methods, supervised hashing seeks compact binary representation of t... more Among learning based hashing methods, supervised hashing seeks compact binary representation of the training data to preserve semantic similarities. Recent years have witnessed various problem formulations and optimization methods for supervised hashing. Most of them optimize a form of loss function with a regulization term, which can be viewed as a maximum a posterior (MAP) estimation of the hashing codes. However, these approaches are prone to overfitting unless hyperparameters are tuned carefully. To address this problem, we present a novel fully Bayesian treatment for supervised hashing problem, named Bayesian Supervised Hashing (BSH), in which hyperparameters are automatically tuned during optimization. Additionally, by utilizing automatic relevance determination (ARD), we can figure out relative discriminating ability of different hashing bits and select most informative bits among them. Experimental results on three real-world image datasets with semantic information show that BSH can achieve superior performance over state-of-the-art methods with comparable training time.
This is the preface to the 10th Intelligent Environments conference held in Shanghai from the 30t... more This is the preface to the 10th Intelligent Environments conference held in Shanghai from the 30th of June to the 4th of July 2014. Intelligent Environments (IEs) use networked computing technology to create environments that are sensitive and responsive to the presence of people. Intelligent Environments are as varied as the places people inhabit encompassing settings such as homes, classrooms, factories, transport, clothing and online virtual worlds, They are constructed using a mix of innovative software and hardware concepts and, as such, promote the efficiency of transactions and activities, facilitate presence and participation, improve user experience and enable better or new lifestyles or methods of production. The conference provided a multidisciplinary interactive forum for researchers and practitioners to present theoretical and practical results of their work all of which are reported in this proceedings.
Q&A is a very important step of a process of learning. This paper analyses some probl... more Q&A is a very important step of a process of learning. This paper analyses some problems in Q&A which both teachers and learners are facing in e-learning, and proposes a novel Q&A mechanism combining automatic Q&A and collaborative Q&A technologies. In detail, this paper proposes an intelligent answer machine using Semantic Words, Keywords and their respective weights to describe the
Based on the modern student-oriented education mode, the paper investigates the importance of t... more Based on the modern student-oriented education mode, the paper investigates the importance of the assessment system in web-based distance learning education environment and establishes an intelligent assessment system model which is characterized for its positive performance and adaptability. Besides, an implementation of the assessment system is given, based on this model. In an integrated education environment, this kind of assessment system cooperates with other subsystems during teaching and learning processes, and plays an important role in improving the performance of the whole education system.
In this ‘work in progress’ paper, we describe research into e-learning and virtual reality at Sha... more In this ‘work in progress’ paper, we describe research into e-learning and virtual reality at Shanghai Jiaotong University (SJTU) and Essex University respectively. We use our experience to propose a novel virtualised learning model and architecture to support online learners and teachers to visualize important learning related information that has hitherto been invisible to users of online learning systems. The purpose of this work is to enhance the learners’ experience (eg by introducing dynamic learning communities), and to help teachers by providing more efficient feedback tools to improve their teaching (eg by creating the notion of a ‘virtual learner’ to represent a group of similar ability learners). In undertaking this research we address challenges, such as learner community composition, virtual classroom construction, avatar modelling, collaborative Q&A machine interaction, etc and propose some novel solutions that will form our future research and experimental agenda.
Crowd counting is a challenging task in computer vison field and haven’t been well addressed unti... more Crowd counting is a challenging task in computer vison field and haven’t been well addressed until now. In this paper, we intend to develop an end to end multi-scale deep convolutional neural network(CNN) model that can accurately estimate the crowd count from an individual image with arbitrary crowd density and perspective. The proposed model extract multi-scale deep CNN features from the input image and regress the crwod count directly, without any post-processing . Hence our model could handle muti-scale targets well in various crowd scene. We evaluate our model on several benchmark datasets and the performance outperforms some state-of-the-art methods. What’s more, due to the end-to-end characteristics, our model demonstrates good practical application performance.
In this 'work in progress' paper, we describe research into e-learning and virtual realit... more In this 'work in progress' paper, we describe research into e-learning and virtual reality at Shanghai Jiaotong University (SJTU) and Essex University respectively. We use our experience to propose a novel virtualised learning model and architecture to support online learners and teachers to visualize important learning related information that has hitherto been invisible to users of online learning systems. The purpose of this work is to enhance the learners' experience (eg by introducing dynamic learning communities), and to help teachers by providing more efficient feedback tools to improve their teaching (eg by creating the notion of a 'virtual learner' to represent a group of similar ability learners). In undertaking this research we address challenges, such as learner community composition, virtual classroom construction, avatar modelling, collaborative Q&A machine interaction, etc and propose some novel solutions that will form our future research and expe...
Among learning based hashing methods, supervised hashing seeks compact binary representation of t... more Among learning based hashing methods, supervised hashing seeks compact binary representation of the training data to preserve semantic similarities. Recent years have witnessed various problem formulations and optimization methods for supervised hashing. Most of them optimize a form of loss function with a regulization term, which can be viewed as a maximum a posterior (MAP) estimation of the hashing codes. However, these approaches are prone to overfitting unless hyperparameters are tuned carefully. To address this problem, we present a novel fully Bayesian treatment for supervised hashing problem, named Bayesian Supervised Hashing (BSH), in which hyperparameters are automatically tuned during optimization. Additionally, by utilizing automatic relevance determination (ARD), we can figure out relative discriminating ability of different hashing bits and select most informative bits among them. Experimental results on three real-world image datasets with semantic information show that BSH can achieve superior performance over state-of-the-art methods with comparable training time.
This is the preface to the 10th Intelligent Environments conference held in Shanghai from the 30t... more This is the preface to the 10th Intelligent Environments conference held in Shanghai from the 30th of June to the 4th of July 2014. Intelligent Environments (IEs) use networked computing technology to create environments that are sensitive and responsive to the presence of people. Intelligent Environments are as varied as the places people inhabit encompassing settings such as homes, classrooms, factories, transport, clothing and online virtual worlds, They are constructed using a mix of innovative software and hardware concepts and, as such, promote the efficiency of transactions and activities, facilitate presence and participation, improve user experience and enable better or new lifestyles or methods of production. The conference provided a multidisciplinary interactive forum for researchers and practitioners to present theoretical and practical results of their work all of which are reported in this proceedings.
Q&A is a very important step of a process of learning. This paper analyses some probl... more Q&A is a very important step of a process of learning. This paper analyses some problems in Q&A which both teachers and learners are facing in e-learning, and proposes a novel Q&A mechanism combining automatic Q&A and collaborative Q&A technologies. In detail, this paper proposes an intelligent answer machine using Semantic Words, Keywords and their respective weights to describe the
Based on the modern student-oriented education mode, the paper investigates the importance of t... more Based on the modern student-oriented education mode, the paper investigates the importance of the assessment system in web-based distance learning education environment and establishes an intelligent assessment system model which is characterized for its positive performance and adaptability. Besides, an implementation of the assessment system is given, based on this model. In an integrated education environment, this kind of assessment system cooperates with other subsystems during teaching and learning processes, and plays an important role in improving the performance of the whole education system.
In this ‘work in progress’ paper, we describe research into e-learning and virtual reality at Sha... more In this ‘work in progress’ paper, we describe research into e-learning and virtual reality at Shanghai Jiaotong University (SJTU) and Essex University respectively. We use our experience to propose a novel virtualised learning model and architecture to support online learners and teachers to visualize important learning related information that has hitherto been invisible to users of online learning systems. The purpose of this work is to enhance the learners’ experience (eg by introducing dynamic learning communities), and to help teachers by providing more efficient feedback tools to improve their teaching (eg by creating the notion of a ‘virtual learner’ to represent a group of similar ability learners). In undertaking this research we address challenges, such as learner community composition, virtual classroom construction, avatar modelling, collaborative Q&A machine interaction, etc and propose some novel solutions that will form our future research and experimental agenda.
Uploads
Papers by Tongzhen Zhang