Gender classification is a difficult but also an essential task under the researches of pattern recognition. There are several methods and features used for this task such as face, gait, or full body features. One of the most widely used techniques is Haar cascades. Default Haar features based classifiers can only detect pedestrian, free from gender information. In this paper we aimed to learn the gender of the target pedestrians by Haar cascades that are trained gender specific. We trained the classifier with only male and female images as positive and negative respectively. Once a basic pedestrian detection has been made over whole image, second detection is made in ROI (Region of Interest) which is the first detected rectangle. Even though we implemented this idea for only pedestrians in this step, it can be applied to other binary problems.

Abstract.Measurement of the image and video quality is crucial for many aspects,such as transmission, compression, perception.The most of traditional methods learning-based image quality assessment(IQA) build the mapping function of the distortion and mass fraction. However,the mapping function is hard to built,and not accurate enough to show the relationship between the linguistic description and numerical number. In this paper,we proposed a new framework to blindly evaluate the quality of an image by learning the regular pattern from natural scene statistics (NSS).Our framework consists of two stages. Firstly,the distortion image is presented by NSS.The Deep Belief Network (DBNs) is used to classify the NSS features to several distortion types. Secondly,a newly qualitative quality pool is proposed according to the distortion types,which converts the distortion types of the image and the degree of the distortion into the numerical scores.In this paper,he proposed distortion classification method is not only more natural than the regression-based,but also more accurate.The experience is conducted on the LIVE image quality assessment database. Extensive studies confirm the effectiveness and robustness of our framework.

ABSTRATAt present,the desktop cloud has come into people's life.You'll see them in some companies, offices, and school labs.The problem I am trying to solve in this paper is that because the computing power of virtual desktop is provided by the virtual machine,and then transmit data to the virtual desktop to display through the network,so compared with the traditional PC in terms of performance, there are still gaps, it is difficult to cope with the application of high load ,such as 3D animation, high-definition video processing, etc.The approach I adopt to solve the problem is that we can use vfio-pci technology to direct GPU to a virtual machine,Enables the virtual machine to monopolize the graphics card,And be able to achieve more than 90% of the performance of graphics,This is a huge improvement for users, and many desktop cloud users are not limited to general enterprise applications,Such as Office, Web applications, Flash playback, video playback and so on.After pass-through,Users can carry out high load applications, 3D animation, high-definition video processing, etc.There are several companies doing virtualization, but KVM is an open source virtualization solution, its market share is not very high, KVM virtualization based on direct display card is not a mature scheme, this is another contribution to the open source.

For color images in a complex background, we cannot be able to detect faces quickly. So we put forward an algorithm, which is based on skin color feature and the improved AdaBoost algorithm. First, through the skin color detection to excluding large amounts of complex background of non-face, after that define the face candidate regions. Besides, when the image is darkness, we will increase the light treatment, afterwards use AdaBoost algorithm to detect the human face, to improve the accurate rate of face detection system and reduce the error rate. In addition, based on the AdaBoost algorithm of former research ,we add new Haar features and modify the weight of its update method, so under the condition of the less weak classifier, the AdaBoost algorithm's training speed much faster, and to prevent the excessive distribution in the process of the training. The experimental results show the proposed method has great improvement for face detection.

Abstract. In order to achieve high efficiency and low cost moving target tracking in video sequences, we proposed parallel particle filtering target tracking algorithm based on Hadoop cloud platform, the algorithm using the open source calculation model to realize the parallel calculation of all particles in the particle filter.The experiments show that the parallel particle filter target tracking algorithm based on Hadoop improve the calculation efficiency when compared with the existing algorithms improve the calculation efficiency.