62results about How to "Avoid neglect" patented technology

The invention discloses a gesture recognition method based on 3D-CNN and convolution LSTM. The method comprises the steps that the length of a video input into 3D-CNN is normalized through a time jitter policy; the normalized video is used as input to be fed to 3D-CNN to study the short-term temporal-spatial features of a gesture; based on the short-term temporal-spatial features extracted by 3D-CNN, the long-term temporal-spatial features of the gesture are studied through a two-layer convolutional LSTM network to eliminate the influence of complex backgrounds on gesture recognition; the dimension of the extracted long-term temporal-spatial features are reduced through a spatial pyramid pooling layer (SPP layer), and at the same time the extracted multi-scale features are fed into the full-connection layer of the network; and finally, after a latter multi-modal fusion method, forecast results without the network are averaged and fused to acquire a final forecast score. According to the invention, by learning the temporal-spatial features of the gesture simultaneously, the short-term temporal-spatial features and the long-term temporal-spatial features are combined through different networks; the network is trained through a batch normalization method; and the efficiency and accuracy of gesture recognition are improved.

The invention discloses an alarm information processing method and apparatus. The method comprises: presetting alarm levels and initial weights of all alarm information; receiving alarm information; according to the alarm level and the initial weight of the of alarm information, calculating a sending time interval of the alarm information by using a memory model; and sending the alarm information based on the time interval. According to the invention, the occurrence frequency of all kinds of alarm information during alarming can be adjusted dynamically according to the levels of the information in the system, so that system maintenance can be carried out by the staff conveniently.

The invention provides a forecasting method for a customer purchase intention and a terminal device, and is applicable to the technical field of information processing. The method comprises the following steps: obtaining personal characteristic data of a customer; inputting the personal characteristic data into a pre-established random forest model to output an objective purchase intention value of the customer; obtaining the subjective purchase intention value of the customer according to the emotion tendency of the customer in the historical telesales process; weighting the objective purchase intention value and the subjective purchase intention value, and outputting the weighted result as the actual purchase intention degree of the customer; and determining the customer with the actualpurchase intention degree greater than the preset threshold value as a potential customer to enable the telesales agent to make phone calls to potential customers and sell the telesales product. The forecasting method provided by the invention integrates a plurality of consideration factors to determine the potential customers, thereby improving the forecasting accuracy of the potential customers;and the quantitative calculation of the purchase intention of the customer is realized by weighting the objective purchase intention value and the subjective purchase intention value.

The invention relates to a Storm task expansion scheduling algorithm based on data stream prediction, and belongs to the field of data exchange networks. Through a monitoring module, the real-time operation data of a Topology task submitted by a user can be obtained, the degree of parallelism of a connected component in Topology under a situation that a component load is met is solved, and then, the degrees of parallelism of all components in the Topology can be solved through iteration. A time series model is used for predicting a data size which needs to be processed by the Topology, the optimal degree of parallelism of a startup component spout in the Topology under the situation is solved, the optimal degree of parallelism of each component in the Topology under a prediction condition is obtained, and scheduling is carried out. In scheduling, an on-line scheduling algorithm is used to reduce inter-node network communication to a largest degree and guarantee the load balance of a cluster. By use of the algorithm, a deficiency that relevance among all components in the Topology is not fully considered is overcome, the deficiency that the optimal degree of parallelism of each component in the Topology submitted by the user can not be quickly and efficiently solved is mad up, and the algorithm has the advantages that change is predicted in advance, handling capacity is improved and handling time delay is lowered.

The invention discloses a random projection multi-kernel learning-based hand gesture identification method comprising the following steps: hand gesture images are collected and preprocessed, preprocessing operation comprises hand gesture positioning operation and hand gesture segmenting operation, sift characteristics are extracted from preprocessed and segmented hand gestures, a K-means algorithm is adopted for training a learning dictionary, an iteration dictionary is used for updating the algorithm and the dictionary, the gesture images are subjected to space pyramid dividing operation, the trained dictionary is used for encoding the sift characteristics of the hand gesture images in each space pyramid layer, and therefore characteristic vectors can be obtained and subjected to cascading operation; random projection is adopted for subjecting the characteristic vectors to dimensional reducing operation; as for a characteristic vector learning kernel matrix after dimensional reducing of each pyramid layer, a multi-kernel model learning algorithm is adopted for classified learning, and an optimal kernel matrix combination coefficient is obtained. Via the method disclosed in the invention, problems of background interference, high complexity, long time consumption, low identification rate and the like in a conventional hand gesture identification method can be solved.

The invention discloses a textile tearing performance testing device which comprises a workbench, supporting columns are fixedly connected to the periphery of the bottom end of the workbench, a discharging groove is fixedly connected to one side of the top end of the workbench, a through groove is formed in the end, close to the center line of the workbench, of the discharging groove, and a testing mechanism is installed at the top end of the workbench. And an electric slide rail is arranged in the testing mechanism. According to the invention, one end of a to-be-detected textile can be quickly extruded and fixed in a C-shaped plate by rotating a hand-screwed bolt, and the electric slide block can drive a tension meter to slide in the electric slide rail, so that the tension meter can drive the C-shaped plate and one end of the textile to move for a pull test, the operation is simple, time and labor are saved, and the previous tedious fixing mode is changed; and the situation that thetextile falls off in the pulling process can be prevented, and in the moving process of the electric sliding block, a marking needle head can be driven to move at the upper end of a precision ruler, so that a worker observes the pulling distance conveniently.

The invention discloses a filter screen cleanliness detection method and sensor, and air treatment equipment. The filter screen cleanliness detection method includes the steps that a filter screen cleaning signal is acquired; according to the filter screen cleaning signal, the cleanliness detection value of a filter screen within a first preset duration before cleaning is read; according to the cleanliness detection value of the filter screen within the first preset duration before cleaning, a first preset threshold value is updated, wherein the first preset threshold value is the upper limitcorresponding to a cleanliness detection value for generating a filter screen cleaning reminding signal, and the updated first preset threshold value is equal to the cleanliness detection value of thefilter screen within the first preset duration before cleaning, or the updated first preset threshold value is between the non-updated first preset threshold value and the cleanliness detection valueof the filter screen within the first preset duration before cleaning. By means of the technical scheme, the use experience of filter screen cleanliness detection is improved.

The invention relates to an image recognition model generation method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a sample image set; wherein the sample image set comprises a plurality of sample image subsets of which the number of images is sequentially decreased, and the plurality of sample image subsets comprise the same number of image categories; training a to-be-trained image recognition model according to the sample image set to obtain a loss value of the to-be-trained image recognition model; wherein the to-be-trained image recognition model comprises a plurality of branch neural networks; wherein the loss value comprises a target classification loss value and a classification loss value, the target classification loss value is aloss value of the model for the sample image set, and the classification loss value is a loss value of the branch neural network for the corresponding sample image subset; and adjusting model parameters according to the loss value until the loss value is lower than a preset threshold. According to the invention, the image types with a small number of images in the training process can be fully trained, and the effect of image recognition model generation is improved.

The embodiment of the invention provides an image display method and a mobile terminal. The image display method includes: receiving a first input from a user on a display area of the abnormal screenin a case where a target image is displayed on a full screen, wherein the target image includes a first image displayed on the display area and a second image obscured by the non-display area; displaying the second image on the display area in response to the first input. In the embodiment of the invention, through adjustment of the target image displayed on the full screen, the image of the occluded part by the non-display area is displayed on the display area for the user to view, thereby avoiding the user's neglect of the image occluded by the non-display area and reducing the inconveniencecaused by the display blind area of the non-display area.

The invention discloses an auditing method for screening round entry events. The auditing method includes the steps: S1 compressing logs according to preset time granularity and key content classification by a classified compression engine to obtain a compression table; S2 screening class round events in the compression table by a screening auditing engine to obtain non-class round events in the compression table, and acquiring round logs 4A by the non-class round events; S3 decompressing parts of the round logs 4A by a decompression engine according to preset decompression rules; S4 displaying the round logs 4A in a combined manner. Parts of the parts of the round logs 4A do not comfort to other abnormal event auditing rules. By the auditing method, the class round events and the non-class round events can be effectively distinguished, auditing result misjudgment is decreased, mass round events can be rapidly and visually displayed, auditing manpower and time investment is reduced, and auditing efficiency is improved.

The invention discloses a remote health monitoring method for home-based care services. The method comprises the following steps that 1, a portable terminal is worn on the aged and used for detecting body surface health data of the aged; 2, an indoor health examination device is installed and used for detecting body indication data of the aged, and the body indication data of the aged are different from the body surface health data of the aged or comprise the body surface health data of the aged; 3, body surface health data of the aged are acquired; 4, the body surface health data of the aged are judged, the aged is reminded to use the indoor health examination device to perform health examination or inform a medical institution for visiting diagnosis according to a judged result. According to the remote health monitoring method for the home-based care services, the body health of the aged is preliminarily judged by collecting the body surface health data of the aged, when it is discovered that the body health data of the aged are seriously disordered through preliminary judgment, the medical institution is informed in time for the visiting diagnosis, a health examination result of the indoor health examination equipment is not waited, therefore, the illness condition of the aged is prevented from being delayed, and the safety of remote health monitoring on the aged is improved.

The invention discloses a multi-cell cooperative transmission scheme based on interference utilization, and the scheme comprises the steps: configuring M antennas in a base station of each cell, enabling the number of users served in each base station to be K, and configuring single antennas for the served users; the two adjacent base stations send information to all the service users in the two cells at the same time; secondly, respectively calculating measurement scales of interference signals which are received by users in a cell and are transmitted by a base station in the cell and are leaked to users served by an adjacent cell by the base station in the cell by adopting a symbol-level pre-coding mode; designing a precoding vector of the base station to maximize the sum of a minimum measurement scale of a service user receiving signal and a minimum measurement scale of a leakage signal in a cell of the base station; and respectively solving the optimal precoding vectors of two adjacent cell base stations by adopting a distributed processing method, and determining a precoding scheme. The signal quality of the signal received by the overall receiving end is improved, the averagesymbol error rate of the receiving end is reduced, and the throughput of the system is improved.

The invention relates to an intelligent door system and an operation method thereof. The intelligent door system comprises an outer display screen embedded at the outer side of a door, an inner display screen at the inner side of the door, a camera, an intelligent door lock and an injection device. The intelligent door system is characterized by further comprising a controller which is in signal connection with the outer display screen and the inner display screen, wherein a face recognition module, an emotion monitoring module, a voice module, a call module, a video module, an alarm module and a wireless module are arranged in the controller. By arranging the face recognition module, the emotion monitoring module, the voice module, the call module, the video module, the alarm module and the wireless module, the intelligent door system can directly recognize an image captured by the camera, so that a proprietor can directly enter the room without a key; information of a visitor can be known by the proprietor at any moment, thereby preventing the visitor from being neglected; and information of a strange visitor can also be known by the proprietor at any moment, and when the strange visitor plans an illegal activity, an anti-theft measure can be taken to avoid unnecessary loss.

A risk analysis support device that performs risk analysis with a short analysis time includes: a control structure diagram input unit which receives an input of a risk analysis target represented in a form of a control structure diagram that represents the risk analysis target by a block and a control showing a relationship between a block executing control and a passive block controlled by the block. A control loop search is performed in a database for a similar control structure diagram including a subset of controls that matches or is similar to a subset of controls extracted from the control structure diagram input by the input unit and acquires from the database a hazard scenario including a node having a correspondence relationship with a subset of controls included in the similar control structure diagram. A recommendation output unit outputs the acquired hazard scenario in a tree structure.

The invention provides a method for setting labels, a server and a label setting system. In the method, a label display number threshold value and a last label hiding value are set, the display widthof one label is obtained according to the width of a label display region, then labels to be displayed are evenly arrayed in the label display region, and it is guaranteed that the last label can be exposed partially when the number of the labels is larger than the label display threshold value. In this way, a user is reminded of follow-up labels and can move to a next web page for display.

A notification control unit includes a notification setting information selection unit, a notification setting information storage unit, a notification setting information setting unit, and the notification execution unit. Notification setting information is associated with individual setting information to be preserved in the notification setting information storage unit. The notification execution unit performs a notification according to setting of use-notification setting information which is selected by using the individual setting information by the notification setting information selection unit. The created or changed notification setting information is preserved in the notification setting information storage unit.

The invention discloses a method and device for discovering potential high-value passengers. Based on processing of large-scale passenger travel data (historical passenger ticket booking information),travel intentions hidden behind passenger travel are discovered, travel intention distribution of each passenger is obtained, and potential high-value passengers of a reserved airline company are further determined based on the travel intention distribution information of each passenger. The invention provides a potential high-value passenger discovery scheme based on a passenger travel intention. The passenger travel intention can reflect the potential requirements of passengers on airlines. Correspondingly, by mining the travel intention distribution information hidden behind the travel data of the passengers, the potential value of the passengers can be determined more accurately, the potential high-value passengers with fewer current passenger records can be found, and neglect of a traditional passenger value measurement method on the potential high-value passengers is avoided.

According to an embodiment of the present invention, a method of controlling wash timing using an intelligent washer comprises determining to perform a wash, calculating a required time for the wash, predicting a return time of the user, and determining a start time of the wash to terminate the wash corresponding to the return time. According to an embodiment, the washer may be related to artificial intelligence (AI) modules, unmanned aerial vehicles (UAVs), robots, augmented reality (AR) devices, virtual reality (VR) devices, and 5G service-related devices.

The invention discloses an electronic device which comprises a shell and a temperature detecting device and further comprises a controller arranged in the shell and a vibration motor in communicating connection with the controller. The controller can control vibration of the vibration motor according to detected temperature information. Tactile sense type information feedback about temperature can be provided through vibration, even if a visual information feedback mode is in a closed state, a temperature state can be also responded in real time, information feedback can be prevented from being neglected, and the information feedback capability can be improved.

The invention discloses an Internet of Things data collection method based on double cache region AoI perception, and the method specifically comprises the following steps: S1, proposing a data storage and transmission strategy based on double cache regions, and designing a system model and sample storage and transmission processes of the two cache regions; s2, after the strategy in the step S1 is put forward, constructing a double-cache AoI model and an average weighted AoI optimization target under the data transmission scene of the Internet of Things; and S3, designing a selection function according to a result obtained in the step S2, and proposing an algorithm based on a greedy strategy to solve the model so as to complete scheduling of source node samples and collection of data. According to the method, the average weighted AoI is reduced by using the greedy strategy based on the double-cache AoI model, the source node and the new and old samples on the source node are flexibly selected according to the real-time conditions of the link and the source node, fresh data can be provided for the Internet of Things user as much as possible even under the condition that the link resource or the source node energy is limited, and the user experience quality is improved.

The invention relates to a method for controlling a display (4) of a tachograph (DTCO) which can be connected to data memories (7) that are associated with defined groups. The aim of the invention is to improve this method so as to allow for an ergonomic control of the complex functionality of a digital tachograph while requiring little space. For this purpose, individual groups are associated with defined functional groups or dialogs of the display (4) and the operational controls (3). The tachograph (DTCO) identifies the group the data memory (7) is associated with and then displays a selection of functional groups that can be selected by means of operational controls (3). The display (4) displays only functions and functional groups that are associated with the group of the data memory (7).

The invention relates to a content item searching method and device, electronic equipment and a storage medium, and belongs to the technical field of computers. According to the technical scheme provided by the embodiment of the invention, recall is carried out based on the search text after the search request is received, the recalled candidate content items are sorted according to different sorting conditions, and the candidate content items screened out under different sorting conditions are respectively placed in different sorting queues. According to the method, when the search result is finally determined, the search result is determined from the multiple sorting queues, it is guaranteed that all the related content items under all the sorting conditions can be selected as the final search result, and therefore the phenomenon that some content items strongly related to the search text are ignored is avoided, and the search effect is improved.

The invention discloses a device for measuring optical path loss of an optical fiber gyroscope. The device comprises a case, a transformer, a light source plate, a digital display voltmeter and a digital display ammeter, the transformer, the light source plate, the digital display voltmeter and the digital display ammeter are arranged in the case; the IC locking seat and the connector socket are arranged on a side cover of the case; one end of the transformer is connected with an external power supply, and the other end of the transformer is divided into two paths, one path is connected with the IC locking seat, and the other path is sequentially connected with the light source plate, the digital ammeter and the connector socket; the digital display voltmeter is connected to the IC lockingseat in parallel; and the gyroscope to be tested is arranged outside the case. The measurement efficiency is much higher than that of a traditional optical path measurement method and is improved bymore than eight times; a light path is not touched, and introduction of human interference is avoided; and according to the invention, neglect of insertion loss in a traditional optical path measurement method is avoided, and the accuracy is higher.

The invention provides an unbalanced feature selection method based on global minimum redundancy, which comprises the following steps of: firstly, establishing a regularization form SIR of an intra-class divergence matrix which emphasizes a minority class and improves a redundancy measurement method on the form of a general LDA divergence matrix according to the characteristics of unbalanced data; secondly, using the regularization form SIR and the absolute inter-class distance vector s as an intra-class divergence matrix and an input feature score vector respectively and introduced into a GRM model, and obtaining an objective function of a GRM-DFS algorithm; and finally, solving the optimization problem through an augmented Lagrangian multiplier method and a segmented root seeking method. According to the method, the problem of solving the global minimum redundancy feature subset biased to a minority class in the unbalanced data is effectively solved. For a subsequent classification algorithm, the GRM-DFS algorithm is helpful for avoiding overfitting and improving the algorithm performance, so that the knowledge discovery efficiency is improved.

The invention discloses an Internet of Things security analysis method based on sensing and calculation linkage, which aims at the phenomenon of linkage of a sensor and an artificial intelligence algorithm widely existing in the Internet of Things, and analyzes the overall vulnerability of an Internet of Things system in a mode of constructing and injecting an adversarial sample aiming at the artificial intelligence algorithm from a physical domain. The method comprises the following steps: firstly summarizing an Internet of Things system into a plurality of information channels according to front and back links of information processing of the Internet of Things, then carrying out vulnerability analysis on each information channel, establishing an actual test platform for signal injection, judging whether actual threats exist or not, and giving the actual threats to professionals to evaluate the harm degree of the actual threats, and finally, integrating the analysis result of each vulnerability, and evaluating the overall safety degree of the intelligent networking system. According to the method, the potential vulnerability can be comprehensively analyzed, and the safety problem of the system can be more quickly mined.

A method for indicating differential signal lines in a layout is provided, which comprises searching line groups having similar line names at first; dividing the line groups into a first line and a second line; confirming that the first and second lines are connected to a same circuit element; highlighting the first and second lines, thereby eliminating the trouble in searching differential signal lines.

The present application relates to an image recognition model generation method, device, computer equipment, and storage medium, including: obtaining a sample image set; the sample image set includes a plurality of sample image subsets in which the number of images decreases successively, and the plurality of sample image subsets all include The same number of image categories; according to the sample image set, the image recognition model to be trained is trained to obtain the loss value of the image recognition model to be trained; the image recognition model to be trained includes multiple branch neural networks; the loss value includes the target classification loss value and classification loss value, the target classification loss value is the loss value of the model for the sample image set, and the classification loss value is the loss value of the branch neural network for the corresponding sample image subset; adjust the model parameters according to the loss value until the loss value is lower than preset threshold. The application enables sufficient training of image categories with a small number of images in the training process, and improves the effect of image recognition model generation.