Abstract: The present disclosure discloses an Internet-of-Things-oriented machine learning container image download system and a method. The Internet-of-Things-oriented machine learning container image download system includes a master node and a plurality of computing nodes; the master node is configured to store and convert a machine learning model, and build a machine learning container image from the format-converted machine learning model; and issue an image download instruction to each of the computing nodes after image information of the machine learning container image is completely built; and each of the computing nodes is configured to receive the image download instruction, download the machine learning container image, and start a machine learning container; and receive data collected by Internet-of-Things devices, and return a data processing result to the Internet-of-Things devices.

Abstract: The present application discloses a task scheduling method based on an improved particle swarm optimization algorithm, which includes: obtaining task data to be scheduled, encoding particles according to the task data; iterating the particles by a particle swarm optimization algorithm; in response to that the particle swarm optimization algorithm does not fall into a local optimal solution, outputting a scheduling scheme; and in response to that the particle swarm optimization algorithm falls into the local optimal solution, outputting the scheduling scheme by fusing the particle swarm optimization algorithm with a cuckoo search algorithm. The present application introduces a cuckoo search algorithm when the particle swarm optimization algorithm falls into a local optimal solution, solving the dilemma of the particle swarm optimization algorithm falling into a local optimal solution, while improving the global search capability of the algorithm.

Abstract: An image defogging method based on dark channel prior is provided, including: acquiring a dark channel image and a light channel image of a haze weather image according to a haze weather image; selecting pixel values in the light channel image and calculating an atmospheric light value A; obtaining a transmittance image based on the light channel image and the atmospheric light value A; obtaining a restored image of defogging based on the transmittance image and the atmospheric light value A. By using the pixel values of the light channel to distinguish whether pixel points comply with the dark channel prior, calculating the transmittance of image areas that comply with the dark channel prior and those that do not, so as to avoid the situation where the image defogging effect is not good due to inaccurate transmittance estimation, making the defogging of images containing high brightness areas more clear.

Abstract: An indoor target positioning method based on an improved convolutional neural network (CNN) model includes acquiring and preprocessing target camera serial interface (CSI) data of a to-be-positioned target and matching the preprocessed target CSI data with fingerprints in a positioning fingerprint database to obtain coordinate information of the to-be-positioned target. The generation method of the positioning fingerprint database includes: collecting indoor WiFi signals by a software defined radio (SDR) platform to obtain indoor CSI data corresponding to the WiFi signals, and preprocessing the indoor CSI data; partitioning the preprocessed indoor CSI data into a plurality of data subsets through a clustering algorithm; training an improved CNN model by the data subsets to obtain a trained improved CNN model; and generating the positioning fingerprint database by the trained improved CNN model and the preprocessed indoor CSI data.

Abstract: Disclosed is a multi-scale fusion defogging method based on a stacked hourglass network, including inputting a foggy image into a preset image defogging network; and outputting a fogless image after the foggy image is processed by the image defogging network. The image defogging network includes a 7×7 convolutional layer, a stacked hourglass module, a feature fusion, a multi-scale jump connection module, a 1×1 convolutional layer, a 3×3 convolutional layer, a hierarchical attention distillation module, the 3×3 convolutional layer and the 1×1 convolutional layer connected sequentially.

Abstract: A method for small object detection in drone scene based on deep learning is provided, which includes: inputting images captured by a drone into a pre-trained generator based on an Unet network structure to output normal-light images; inputting the normal-light images into a object detection backbone network to output a plurality of multidimensional matrix feature maps, wherein the object detection backbone network integrates a channel attention mechanism and a spatial attention mechanism based on convolutional block Self-Block, and a 7*7 large convolutional kernel is used; inputting the plurality of multidimensional matrix feature maps into a BiFPN-S module of a feature pyramid for feature fusion, so as to output a plurality of corresponding feature maps for predicting objects of different sizes.

Abstract: A container image management system for distributed clusters, the system including at least one master node and at least one worker node. The at least one master node includes a container image database, a request input module and a container image management module. The container image management module is responsive when the container image management module establishes the connection to the container image database, then it is configured to perform a read/write operation on the container image database. The container image database is a distributed database configured to store node information of the at least one master node and the at least one worker node in the container image management system. The request input module is configured to receive request content including a request destination and command execution content. The command execution content includes an execution operation field and an executed container image list.

Abstract: The present invention pertains to the technical field of smoke detection and discloses a lightweight fire smoke detection method, a terminal device, and a storage medium. Primarily, a smoke-like foreground is screened out based on a combination of a mixed Gaussian model and a YUV color model; and an ACON activation function is introduced to replace a Leaky ReLU activation function of YOLOv5 to form an ACON-CSP module for feature extraction. In the present invention, the smoke foreground is extracted for preprocessing by using the combination of the mixed Gaussian and YUV. Thus, in the preprocessing stage, static interference and non-smoke color interference in the preprocessing stage are ruled out while narrowing down the range of smoke detection, ensuring the relative accuracy and improving the detection speed, thereby providing a possible method for fire smoke detection of a low-end terminal device in an outdoor computer.

Abstract: The present invention discloses a method and an apparatus for obstacle detection under complex weather. The method includes: obtaining an image under a complex weather condition; performing enhanced preprocessing on the image by using a multi-scale retinex with color restoration MSRCR algorithm; inputting the preprocessed image into a trained obstacle detection model based on an improved YOLOv3 network; and according to output of the obstacle detection model based on the improved YOLOv3 network, determining an obstacle detection result under the complex weather; replacing a Leaky-ReLU activation function in convolutional layers in the original YOLOv3 network with an ELU activation function; and training the obstacle detection model with the processed data set to obtain a trained obstacle detection model based on the improved YOLOv3 network.

Abstract: The present invention discloses an instrument recognition method based on an improved U2 network. The method includes: replacing common convolution of each layer with grouped convolution Grouped Cony on a basis of an RSU, and segmenting a dial plate and a pointer by using the network; performing noise reduction processing on the scale value array by using a mean filter; and determining a position of a scale value corresponding to the pointer by using a peak value, and outputting a reading according to the scale value and preset data. A pointer-type instrument is segmented by using the improved U2 network, and automatic reading of the obtained dial plate is implemented by using a conventional computer vision method. Therefore, compared with manual reading, the method has advantages of high precision, high reliability, fast reading, low costs, and the like, and working efficiency can be greatly improved.

Abstract: The present disclosure discloses an image segmentation method combined with superpixel and multi-scale hierarchical feature recognition. This method is based on a convolutional neural network model taking multi-scale hierarchical features extracted from a Gaussian pyramid of an image as a recognition basis, and then being connected with a multilayer perceptron to achieve the recognition of each pixel in the image, moreover, this method is used tier performing superpixel segmentation on the image and is combined with a method for improving superpxiel in combination with LBP texture features to segment an original image so that an obtained superpixel block is more fitted to edges of targets, then, the original image is merged according to a mean value of a color, and finally, recognition of each target in the image is achieved.

Abstract: The disclosure provides a method for detecting and defending a Distributed Denial of Service attack in an SDN environment. The method includes: building data messages acquired as feature messages by a proxy module; sending the feature messages to a pre-built detection model to obtain a detection result; making a decision instruction based on the detection result; and performing control operations by the proxy module based on the decision instruction.

Abstract: Disclosed are an indoor passive human behavior recognition method and device. The method includes the following steps: dividing an indoor activity space into multiple regions, collecting a channel impulse response data packet of a reflection signal of each activity in each region to obtain an H (M, N, Z) matrix; preprocessing the H (M, N, Z) matrix to obtain a preprocessed H (M, N, Z) matrix; extracting features of the preprocessed H (M, N, Z) matrix to obtain a training sample of a convolutional neural network model; performing transfer learning on the convolutional neural network model using the training sample to obtain a trained convolutional neural network model; obtaining an indoor channel impulse response amplitude value, inputting the channel impulse response amplitude value into the trained convolutional neural network model, and outputting a human behavior.

Abstract: A method for recognizing a fog concentration of a hazy image includes inputting a target hazy image into a pre-trained directed acyclic graph (DAG) support vector machine to acquire a fog concentration of the target hazy image. The fog concentration of the target hazy image is represented based on a prebuilt multi-feature model, and the feature vector in the multi-feature model includes at least one of a color feature, a dark channel feature, an information quantity feature and a contrast feature.

Abstract: A method for detecting comprehensive GPU-related factors of a distributed cluster, the method including: (1): checking whether there is a configuration file content of an operating node; (2): reading a mode parameter in an environment variable of the operating node, and correspondingly switching an operating mode according to the mode parameter; (3): reading a timer frequency value from the environment variable of the operating node so as to set a time period for reading a GPU information parameter according to the timer frequency value; (4): calculating the maximum value of the GPU information parameter of the operating node, and storing the maximum value into the GPU information list cache; and (5): initializing the transmitted information; determining whether there is a GPU in the GPU information list cache of the operating node.

Abstract: The present disclosure discloses a multiscale point cloud classification method. The method includes the following steps: acquiring 3D unordered point cloud data; performing feature extraction and classification on the acquired point cloud data using a pre-trained parallel classification network to obtain an output result, wherein the parallel classification network includes a plurality of basic networks with the same structures; and fusing the output results of the parallel network using a pre-trained deep Q network to obtain a final result of point cloud classification. The present disclosure can improve the accuracy and robustness of point cloud classification.

Abstract: The present disclosure discloses a high-precision point cloud completion method based on deep learning and a device thereof, which comprises the following steps: introducing dynamic kernel convolution PAConv into a feature extraction module, learning a weight coefficient according to the positional relationship between each point and its neighboring points, and adaptively constructing the convolution kernel in combination with the weight matrix. A spatial attention mechanism is added to a feature fusion module, which facilitates a decoder to better learn the relationship among various features, and thus better represent the feature information. A discriminator module comprises global and local attention discriminator modules, which use multi-layer full connection to classify and determine whether the generated results conform to the real point cloud distribution globally and locally, respectively, so as to optimize the generated results.

Abstract: The present disclosure discloses an image haze removal method and apparatus, and a device. The method includes: acquiring a hazy image to be processed; and obtaining a haze-free image corresponding to the hazy image by inputting the hazy image into a pre-trained haze removal model. The present disclosure uses the residual dual attention fusion modules as basic modules of the neural network, so that each feature map can obtain pixel features while enhancing the global dependence, thus improving the image dehazing effect.

Abstract: Disclosed in the present invention are a distributed container image construction scheduling system and method. The system includes a construction node and a management node. The construction node includes an image constructor for executing a construction task issued by the management node. The management node includes a console and a scheduler. The console is responsible for acquiring the relevant parameters such as a development dependency library and system configuration required by a user, and generating tasks with these parameters and sending same to the scheduler. The scheduler is used for receiving the tasks sent by the console, detecting the legitimacy of the tasks, and sending the tasks to the corresponding construction node to be run.

Abstract: The present disclosure discloses an image haze removal method and apparatus, and a device. The method includes: acquiring a hazy image to be processed; and obtaining a haze-free image corresponding to the hazy image by inputting the hazy image into a pre-trained haze removal model. The present disclosure uses the residual dual attention fusion modules as basic modules of the neural network, so that each feature map can obtain pixel features while enhancing the global dependence, thus improving the image dehazing effect.