Abstract: A method is provided for webpage pre-reading. The method includes obtaining visiting information during a period of time for multiple webpages of each user from a plurality of users including a current user; determining respective similarity values between the current user and the plurality of users based on the visiting information; determining one or more most-similar users based on the similarity values; and when the current user is browsing a current webpage, predicting a target webpage to be accessed by the most-similar users as a target webpage for the current user and pre-reading the target webpage.

Abstract: The present application discloses a 3D imaging method and apparatus, which divide a source 3D image light into two polarized light beams carrying image information through a birefringence effect, adjust optical paths of the two polarized light beams obtained through birefringence, and control the two polarized light beams to be alternately irradiated, thereby achieving 3D imaging; since the implementation of the above imaging method only requires one imaging apparatus having corresponding functions, the present application can simplify the system structure and reduce the system cost relative to a conventional 3D projection technology that requires two imaging devices; moreover, a viewer can directly see, with no need to wear corresponding 3D glasses, a 3D image due to the parallax caused by alternate irradiation of two polarized light beams carrying image information out of a projection device.

Abstract: Embodiments of the present disclosure disclose a method and a device for webpage preloading. The method includes: conducting webpage preloading according to a current preloading policy, in which the preloading policy includes: a preloading time range, a preloading region, a preloading page depth, and an available caching space for preloading; counting historical data within a pre-set time period, in which the historical data includes: information about an accessed webpage, information about a preload webpage, and state information of a local cache; and updating the preloading policy based on the historical data. In the present disclosure, by way of counting the preloading historical data within a pre-set time period, and based on the changes in the historical data, the preloading policy is automatically updated, so that the preloading policy can adapt to network and user access conditions in real time, thereby improving the hit accuracy of webpage preloading.

Abstract: A single-point dual sensor-based leakage method and system for a gas-liquid stratified flow pipeline includes a single-point dual sensor-based leakage formula for a pipeline to be detected; installing two sensors at the same point of one end of the pipeline, a signal collected by the bottom sensor being an acoustic wave propagated by liquid in the pipeline, and a signal collected by the top sensor being an acoustic wave propagated by gas in the pipeline; processing the two acoustic waves to obtain a time difference; and substituting the acoustic velocity in a gas and the time difference into the single-point dual sensor-based leakage formula. The installation of sensors at two ends of a pipeline, avoids missing detection of leakage acoustic wave signals by a single sensor, reduces the number of installed sensors, and is low in cost, high in safety and strong in applicability to a gas-liquid stratified flow pipeline.

Abstract: Embodiments described herein provide various examples of a joint face-detection and head-pose-angle-estimation system based on using a small-scale hardware CNN module such as the built-in CNN module in HiSilicon Hi3519 system-on-chip. In some embodiments, the disclosed joint face-detection and head-pose-angle-estimation system is configured to jointly perform multiple tasks of detecting most or all faces in a sequence of video frames, generating pose-angle estimations for the detected faces, tracking detected faces of a same person across the sequence of video frames, and generating “best-pose” estimation for the person being tracked. The disclosed joint face-detection and pose-angle-estimation system can be implemented on resource-limited embedded systems such as smart camera systems that are only integrated with one or more small-scale CNN modules.

Abstract: Provided are a window switching method, and device and storage medium thereof. The method comprises the steps: obtaining a first pre-reading keyword meeting the requirement in all contents on a page opened through a current window, determining a first pre-reading URL corresponding to the first pre-reading keyword, utilizing the pre-reading window to load a pre-reading page corresponding to the first pre-reading URL, and switching the pre-reading window loaded with the pre-reading page to the current window when an instruction of entering the pre-reading page is received. An improved pre-reading function can be achieved, and the technical problem of low user experience caused by the fact that contents on a pre-reading page cannot be displayed completely after switching to the pre-reading page is further solved.

Abstract: Embodiments described herein provide examples of a real-time visual object tracking system. In one aspect, an UAV capable of performing real-time visual tracking of a moving object includes a processor; a memory coupled to the processor; and a camera coupled to the processor and the memory and to capture a video of the moving object. This UAV additionally includes a visual tracking module to: receive a first video image and a first location of the object; receive a second video image following the first video image; place a first search window in the first video image and a second search window in the second video image at the same location as the first search window; compute a correlation between an image patch within the first search window and an image patch within the second search window; and determine an updated location of the object in the second video image.

Abstract: Embodiments described herein provide various examples of an automatic obstacle avoidance system for unmanned vehicles using embedded stereo vision techniques. In one aspect, an UAV capable of performing autonomous obstacle detection and avoidance is disclosed. This UAV includes: a stereo vision camera set coupled to the one or more processors and the memory to capture a sequence of stereo images; and a stereo vision module configured to: receive a pair of stereo images captured by a pair of stereo vision cameras; perform a border cropping operation on the pair of stereo images to obtain a pair of cropped stereo images; perform a subsampling operation on the pair of cropped stereo images to obtain a pair of subsampled stereo images; and perform a dense stereo matching operation on the pair of subsampled stereo images to generate a dense three-dimensional (3D) point map of a space corresponding to the pair of stereo images.

Abstract: The present invention provides a webpage pre-reading method, apparatus and a smart terminal device. The webpage pre-reading method includes: acquiring information of one or more users accessing multiple webpages within a first period of time; determining click paths through each of which a second webpage is clicked and entered from a first webpage and the number of clicks thereof; determining a first confidence value of each click path; generating a first pre-reading list including the click paths based on the first confidence values; and querying for a click path of a currently browsed webpage from the first pre-reading list so as to pre-read data of a corresponding target webpage to be browsed.

Abstract: A set of nosepieces are configured to attach to a rivet setting tool. The set of nosepieces comprises a first nosepiece with a longitudinal bore having a first diameter and an O-ring having a first color, and a second nosepiece with a longitudinal bore having a second diameter that is different than the first diameter and an O-ring having a second color that is different than the first color.

Abstract: Embodiments of a convolutional neural network (CNN) system based on using resolution-limited small-scale CNN modules are disclosed. In some embodiments, a CNN system includes: a receiving module for receiving an input image of a first image size, the receiving module can be used to partition the input image into a set of subimages of a second image size; a first processing stage that includes a first hardware CNN module configured with a maximum input image size, the first hardware CNN module is configured to sequentially receive the set of subimages and sequentially process the received subimages to generate a set of outputs; a merging module for merging the sets of outputs into a set of merged feature maps; and a second processing stage for receiving the set of feature maps and processing the set of feature maps to generate an output including at least one prediction on the input image.

Abstract: According to the present disclosure, a liposomal formulation comprising at least one uncharged phospholipid without cholesterol and at least one pharmaceutical agent, wherein the at least one uncharged phospholipid forms one or more lipid bilayers without cholesterol encapsulating the at least one pharmaceutical agent, is provided. The use(s) of such a liposomal formulation and method of producing such a liposomal formulation are also provided.

Abstract: Embodiments described herein provide various examples of a face-image training data preparation system for performing large-scale face-image training data acquisition, pre-processing, cleaning, balancing, and post-processing. The disclosed training data preparation system can collect a very large set of loosely-labeled images of different people from the public domain, and then generate a raw training dataset including a set of incorrectly-labeled face images. The disclosed training data preparation system can then perform cleaning and balancing operations on the raw training dataset to generate a high-quality face-image training dataset free of the incorrectly-labeled face images. The processed high-quality face-image training dataset can be subsequently used to train deep-neural-network-based face recognition systems to achieve high performance in various face recognition applications.

Abstract: A webpage display method includes: acquiring source code of a webpage; parsing the source code to form tree-shaped nodes; composing the tree-shaped nodes; determining a primary container according to the composed tree-shaped nodes; determining at least one secondary container in the primary container; adjusting the area of the secondary container; and drawing the adjusted secondary container and displaying the elements of the adjusted secondary containers on the screen.

Abstract: Disclosed are a method, device, and server for producing a video frame set. The method comprises: compiling statistics on the degree of playback intensity of each video frame in a video, selecting at least one “selected video frame” from the video on a basis of the degree of playback intensity, where the degree of playback intensity of the video frame is a ratio between the number of times that the video frame is played back and the number of times the video is played back; and then, splicing the “selected video frames” to form a video frame set. The present embodiment selects the “selected video frames” on a basis of the degree of playback intensity of the video frames in the video, and splices together the video frames so selected to form a video frame set, thus allowing video contents considered to be highlights by a majority of users to be spliced into the video frame set.

Abstract: This invention comes up with a kind of Android malicious code detection method on the base of community structure analysis. During the reverse analysis process of target program, firstly, it obtains critical static feature information automatically, such as permission, function, class, system API, etc.; secondly, it uses the call relation between functions to create function call graph, and undertakes pretreatment on function call graph; make cycle division and analysis for the weighted function call graph so as to get the correction division of community structure; finally, it extract features from community structures for machine learning and get the final maliciousness determination result. This invention method is able to undertake program internal structure analysis and malicious code detection rapidly when facing a large number of Android application program samples generated by “repackaging” technology.

Abstract: Techniques for performing marking and reminding on contents in a web page are disclosed. A browser client of a mobile terminal presets reminder content and a display style and stores the same in the mobile terminal. The method comprises: acquiring a source code of a web page; looking up the source code to obtain content to be reminded semantically associated with the preset reminder content; and marking and displaying the content to be reminded using the preset display style in the web page. The present invention marks content which concerns a user before a web page is displayed to the user after loading has finished, and marks and displays new content in a designated format using a special preset style, so that the new content is more conspicuous in a page.

Abstract: A method and device are disclosed for presenting content data (such as a video, an audio and a Flash) from a network. In content-related information (such as a webpage file) of related content data from a network, a plurality of content data sources are often provided for the same content. A weighted value is respectively set for each content data source based on the content-related information. Based on the weighted value of each content data source, a content data source is selected so as to present the content data. The weighted values can be set by considering various factors. In an embodiment, in consideration of different supporting capacities of a terminal system for different types of content data, the weighted values are set according to an MIME type of the content data sources.

Abstract: Provided are a method and an apparatus for implementing communication from a web page to a client. The method includes: a browser establishing a communication connection with a port when accessing a web page, where the port is a port predefined to communicate with the web page; and sending link information carrying parameters specified by a preset communication protocol to the port, so that the client accesses a client page corresponding to the link information, where the parameters include fields in a service requirement and client page information. A client can be directly called from a web page of a portal web site, so as to guide a user to complete a required service to a designated page, thereby being convenient for the user to operate.

Abstract: This invention public a kind of malicious software clustering method expressed based on TLSH feature, which belongs to the analysis and test area of malicious software. Firstly, the Cuckoo Sandbox is used to analyze the malicious software to acquire three kinds of character string features, which are the static feature of the software, resource assess record during operation and API; And then the character strings are disassembled, filtered and sorted and the TLSH algorithm is used to compress them into three groups of feature values with size of 70 characters; Finally the OPTICS algorithm is utilized to realize the automatic classification on the malicious software family. This invention adopts unsupervised learning methods, which does not need the manual tab for the training in advance. The features which are extracted are compressed and expressed by using the TLSH.