As deep learning technology advances, human fall detection (HFD) leveraging convolutional neural networks (CNNs) has recently garnered significant interest within the research community. However, most existing works ignore cross-frame association of skeleton keypoints and aggregation feature representations. To address this, we first introduce an image preprocessing (IPP) module, which enhances foreground weakens background. Diverging from common practices that employ off-the-shelf detector...

With the vigorous development of artificial intelligence big data era and advent 5G era, amount network information shows a blowout-like growth. As result, accurate query is faced with unprecedented challenges. The image as material reproduction visual perception, has large number retrieval requests all time, but traditional target recognition annotation mainly based on pixel-level supervised learning. In face massive high-quality recognition, it very difficult for users to content...

Spectral clustering algorithm has proved be more effective than most traditional algorithms in finding clusters. However, its high computational complexity limits effect actual application. This paper combines the spectral with MapReduce, through evaluation of sparse matrix eigenvalue and computation distributed cluster, puts forward improvement ideas concrete realization, thus improves speed distinctive algorithm. According to experiment, processing data scale being enlarged, rate is nearly...

A lignin and cellulose-degrading fungal strain Bio-1 was isolated from the soil of Yuelu Mountain in China. It identified as a member genus Cladosporium by 18s rDNA, ITS sequences analysis morphological characters. could hydrolyze at least 35% alkaline 4 days when incubated screening liquid medium. Both laccase cellulase activities be detected fermentation cell homogenate. High extracellular intracellular 3800 5352 U/L were obtained with copper induction, respectively. The production reached...

The recommendation system has been widely used in life, which greatly facilitates people's life. traditional method is mainly to analyze the interaction between users and items. history of items, get only users' preferences for items past. serialization analyzes sequence interacting with objects a recent period time. To consider relevance user's before after behavior, can obtain preference short term. However, emphasizes connection item Ignoring relationship properties objects. In view above...

In this paper, we obtain the existence of multiple positive solutions a boundary value problem for α-order nonlinear integro-differential equations in Banach space by means fixed point index theory completely continuous operators. MSC:26A33, 34B15.

Image segmentation is an important research in image processing and machine vision which automated driving can be seen the main application scene of algorithms. Due to many constraints power supply communication in-vehicle systems, vast majority current algorithms are implemented based on deep learning model. Despite ultrahigh accuracy, problem mesh artifacts being too severe obvious, high cost, computational, consumption devices required difficult apply real-world scenarios. It focus this...

Metamorphic malware is able to change its signature which a fixed sequence of bytes or code from generation generation. This makes the traditional scanner based on difficult detect it. Combing normalization with promising direction solve problem. The idea is: firstly transform into canonical form then pass it check. paper proposes novel method towards instruction substitution commonly used metamorphism approach. mainly involves two steps. Firstly, input rewritten standard instructions....

Original scientific

The video coding perceptual quality at low bit rates was enhanced by a rate control scheme that achieves better trade off between consistent and smooth buffer occupancy. A proportional integral technique used for the frame level control. motion extrapolated complexity propagation model to allocate bits region of interest (ROI)and other regions while smoothing distortion in ROI using optimized macro block allocation. Simulations show algorithm reduces fluctuation 35.5% 57.1% compared with...

Knowledge Distillation (KD) and Model Averaging (MA) represent prominent methods forenhancing the efficiency effectiveness of Deep Neural Networks (DNNs). Averagingis a method that involves producing an average model to approximate ensemble by taking theaverage multiple checkpoints along trajectory stochastic gradient descent (SGD). On theother hand, is technique in which knowledge transferred from ateacher student model. Regrettably, we have found these two promising techniquesare...

Non-intrusive load identification plays a crucial role in developing green and low-carbon energy supply demand mechanism. Among the various technologies, low-frequency electrical signal stands out due to its ability discern user's electricity consumption habits without increasing hardware cost, thus presenting promising application prospects. However, challenge lies scarcity of available data uneven distribution samples, resulting reduced accuracy. In this paper, we present non-intrusive...

Macvlan and IPvlan modes allow the host physical Network Interface Card (NIC) to create several virtual sub interfaces, which can meet containers communication requirements across different network segment in complex application scenarios. The paper describes architecture characteristics of these two depth builds container overlay single machine multi environments. Around key indicators such as round-trip time (RTT), bandwidth jitter, Ping Iperf tools are used test analyse end-to-end...

Unmanned aerial vehicles (UAVs) are widely used in power transmission line inspection nowadays and they need to navigate automatically by recognizing the category accurate position of pylon equipment inspection. Semantic segmentation is an effective method for equipment. In this paper, a semantic algorithm that fuses infrared natural light images proposed. A cross-modal attention interaction activation mechanism adopted fully exploit complementation between images. Firstly, global...

Image semantic segmentation as a kind of technology has been playing crucial part in intelligent driving, medical image analysis, video surveillance, and AR. However, since the scene needs to infer more semantics from audio clips request for real-time performance becomes stricter, whetherthe single-label classification method that was usually used before or regular manual labeling cannot meet this end. Given excellent deep learning algorithms extensive applications, algorithm based on...

In this era of mobile application and socialization, location-based services (LBSs) have become unprecedentedly popular emphasized. By submitting its location to the service provider, terminal can obtain useful service. As an important technology that provide services, spatial query processing has a research hotspot. According problems existing in current road network partitioning, first all, paper proposes partition density formula optimal partitioning method according density. Based on...

As the third wave of IT following personal computers, Internet, cloud computing is not only revolution modern information technology, but also more innovation business model. From to services, "cloud" has led an industrial revolution, and various services emerge endlessly. The technology service in China rapid development areas across country are advancing corresponding plans. But with continued growth type quantity, users faced challenge how choose best services. Based on analysis service,...