6results about How to "Solving recognition difficulties" patented technology

The invention provides a multi-round dialogue intention classification method and device, and the method comprises the steps: defining and extracting a single-valued slot which is a core keyword contained in an FAQ intention; constructing intention clusters based on the single-valued slots, wherein the intention clusters of the same entity type and the intention clusters of the same single-valued slots are constructed; a single-valued slot inheriting slot mechanism is introduced into multiple rounds of dialogues, and different dynamic score adding strategies are performed on the intention clusters of the same entity type and the intention clusters with the same single-valued slot based on the recognition result that the hit intention has the single-valued slot; and performing intention classification again according to the intention score after score addition. The intention recognition accuracy and effectiveness can be improved.

The application discloses a live working anti-external damage monitoring system and method, relates to the technical field of power operation safety protection, and the system comprises a laser radar, an electric field sensor, a main control module and a strong magnetic force holder. The bottom of the strong magnetic force holder is provided with a strong magnetic force adsorption disc, the top is provided with a mounting platform, the inside is provided with a horizontal shaft motor and a vertical shaft motor, and the side is provided with a data interface and a power supply interface. The laser radar is connected with the strong magnetic force holder through the mounting platform and is used for collecting laser point cloud data. The electric field sensor is connected with the strong magnetic force holder through the mounting platform and is used for collecting electric field fluctuation data. The main control module is connected with the strong magnetic force holder through a shielding cable and is used for acquiring operation area data and performing anti-external damage monitoring according to the operation area data. Through the cooperative matching of the laser radar, the electric field sensor, the main control module and the strong magnetic force holder, effective anti-external damage monitoring of the operation area during live working can be realized, and the safety of live working is improved.

The invention discloses an object recognition method, device and equipment and a medium, and the method comprises the steps: obtaining a shot image, and carrying out the preprocessing of the shot image, and the preprocessing comprises at least one of denoising, image enhancement and cutting; on the basis of a trained feature extractor, feature extraction is carried out on an object to be recognized in the preprocessed shot image, local features and global features are obtained, the local features are obtained through a convolutional neural network of the feature extractor, and the global features are obtained through a Transform network of the feature extractor; carrying out feature fusion on the local features and the global features to obtain a comprehensive feature vector; and carrying out similarity calculation on the comprehensive feature vector and a feature vector of a known object so as to identify the to-be-identified object. According to the object recognition method, the convolutional neural network and the Transform are combined to fuse the advantages of the convolutional neural network and the Transform, so that the image features are extracted more comprehensively and greatly, and the accuracy of object recognition is improved.

The present disclosure relates to the technical field of chemical warehouse management, and discloses a chemical storage method and a chemical storage system. The central area of the bearing table of the storage system is provided with a rotatable object table for placing the chemical to be stored. A moving assembly is connected to the camera assembly. A first display is provided with a first auxiliary strip and a second auxiliary strip on the first preview area of the first display. The first auxiliary strip is a frame-shaped structure coaxially arranged with the first preview area, and the area of the frame-shaped structure is smaller than the area of the first preview area. The second auxiliary strip overlaps the symmetry axis of the first preview area, and the two ends of the second auxiliary strip are connected to the first auxiliary strip. The chemical storage method and the chemical storage system can extract and compare the packaging box and reagent bottle label information, and filter and enter the parameter value from the packaging box and reagent bottle label information according to the logical relationship and weight rules.

The invention discloses a photovoltaic cleaning robot state recognition system, and relates to the technical field of robot control, in particular to the photovoltaic cleaning robot state recognition system. The system comprises a data processing unit which is used for preprocessing collected original data, and the original data is dual-motor working current. The multi-channel convolution-gating loop unit (MCNN-GRU) is used for taking preprocessed data as input, the self-attention unit (Self-attention) is used for further enhancing features obtained after model processing and paying attention to areas with large fluctuation, and the problem that individual state recognition effects are poor can be solved. According to the method, data are processed through the MCNN-GRU-SA model, the current running state of the robot is obtained, reliable running guarantee can be provided for the unattended photovoltaic cleaning robot, and intelligent monitoring and early fault early warning of equipment are facilitated.

A method, electronic device, medium, and computer product for recognizing hand-drawn flowcharts are disclosed. The method includes: acquiring stroke features corresponding to strokes in the hand-drawn flowchart; inputting the stroke features into a graph neural network model to obtain the predicted categories of the strokes in the hand-drawn flowchart output by the graph neural network model, and symbol features obtained by clustering the stroke features, wherein the graph neural network model is trained from sample stroke features; and determining the category of the symbol features based on the predicted categories of the strokes in the hand-drawn flowchart. Implementing the technical solution provided by this invention can improve the recognition accuracy of symbols in hand-drawn flowcharts and effectively solve the problem of difficulty in recognizing hand-drawn flowcharts in the prior art.